Get 20M+ Full-Text Papers For Less Than $1.50/day. Start a 14-Day Trial for You or Your Team.

Learn More →

Mapping myelin in white matter with T1-weighted/T2-weighted maps: discrepancy with histology and other myelin MRI measures

Mapping myelin in white matter with T1-weighted/T2-weighted maps: discrepancy with histology and... The ratio of T1-weighted/T2-weighted magnetic resonance images (T1w/T2w MRI) has been successfully applied at the cortical level since 2011 and is now one of the most used myelin mapping methods. However, no reports have explored the histological validity of T1w/T2w myelin mapping in white matter. Here we compare T1w/T2w with ex vivo postmortem histology and in vivo MRI methods, namely quantitative susceptibility mapping (QSM) and multi-echo T2 myelin water fraction (MWF) mapping techniques. We report a discrepancy between T1w/T2w myelin maps of the human corpus callo- sum and the histology and analyse the putative causes behind such discrepancy. T1w/T2w does not positively correlate with Luxol Fast Blue (LFB)-Optical Density but shows a weak to moderate, yet significant, negative correlation. On the contrary, MWF is strongly and positively correlated with LFB, whereas T1w/T2w and MWF maps are weakly negatively correlated. The discrepancy between T1w/T2w MRI maps, MWF and histological myelin maps suggests caution in using T1w/T2w as a white matter mapping method at the callosal level. While T1w/T2w imaging may correlate with myelin content at the cortical level, it is not a specific method to map myelin density in white matter. Keywords Corpus callosum · Myelin mapping · Neuroanatomy · Neuroimaging · Validation · Quantitative susceptibility mapping · Myelin water fraction Marco Catani and Flavio Dell’Acqua have equally contributed to the paper. * Stefano Sandrone Department of Neurophysics, Max Planck Institute sandrone.stefano@gmail.com for Human Cognitive and Brain Sciences, Leipzig, Germany * Flavio Dell’Acqua Paul Flechsig Institute of Brain Research, University flavio.dellaqua@kcl.ac.uk of Leipzig, Leipzig, Germany Department of Basic and Clinical Neuroscience, Institute Department of Brain Sciences, Faculty of Medicine, of Psychiatry, Psychology & Neuroscience, King’s College Imperial College London, London, UK London, London, UK NatBrainLab, Department of Forensic Clinical Neuropathology, King’s College Hospital NHS and Neurodevelopmental Sciences, Institute of Psychiatry, Foundation Trust, London, UK Psychology & Neuroscience, King’s College London, London, UK Department of Neuroimaging, Institute of Psychiatry, Psychology & Neuroscience, King’s College London, The Sackler Institute for Translational Neurodevelopment, London, UK Institute of Psychiatry, Psychology & Neuroscience, King’s College London, London, UK Developmental Imaging and Biophysics Section, UCL Great Ormond Street Institute of Child Health, London, UK IRCCS SYNLAB SDN S.p.A., Naples, Italy Institute of Biostructures and Bioimaging, National Research Brain Connectivity and Behaviour Laboratory, Sorbonne Council, Naples, Italy Universities, Paris, France Groupe d’Imagerie Neurofonctionnelle, Institut des Maladies Neurodégénératives-UMR 5293, CNRS, CEA University of Bordeaux, Bordeaux, France Vol.:(0123456789) 1 3 Brain Structure and Function white matter tract, but also the most thoroughly examined Introduction using in vivo and post-mortem techniques in animals and humans (Zaidel and Iacoboni 2003). A further advantage Half of the human brain is composed of white matter tis- is that an established geometric subdivision of the cor- sue, which primarily contains myelinated axons of various pus callosum is thought to reflect different cortical pro- diameters (Alberts et al. 2014; Quarles et al. 2006). Myelin jections (Witelson 1989). According to this subdivision, sheath surrounds axons and is essential for the efficient the rostrum and the genu project to the prefrontal cortex, conduction of action potential between neurons (Nieu- whereas the posterior body and the splenium project to the wenhuys et al. 2007; Kandel et al. 2012). Several human sensory-motor and visual areas, respectively (Zaidel and brain diseases are associated with white matter pathology Iacoboni 2003; Catani and Thiebaut de Schotten 2012). (Lazzarini 2003; Schmahmann and Pandya 2006; van der Microscopy studies of the corpus callosum demonstrated Knaap and Valk 2011; Catani and Thiebaut de Schotten a heterogeneous distribution of fibres along the antero- 2012). It is, therefore, crucial to have reliable and valid posterior axis (Aboitiz et al. 1992; Zaidel and Iacoboni methods to map brain myelination in the living brain. 2003; Caminiti et al. 2009, 2013). Unmyelinated fibres Advances in structural imaging include approaches make up 16% of the genu and under 5% of other callosal towards myelin mapping, such as the ratio of T1-weighted/ areas; moreover, the posterior midbody is the region with T2-weighted Magnetic Resonance Imaging (MRI) scans the highest density of large and highly myelinated callosal (Glasser and Van Essen 2011; Glasser et al. 2014), multi fibres (Aboitiz et al. 1992). This permits us to put forward echo-T2 (MacKay et al. 1994; Zhang et al. 2015), T1 maps specific hypotheses about the topographical distribution (Geyer et al. 2011; Geyer and Turner 2013; Bock et al. of the myelin density along the corpus callosum. Hence, 2013), mcDespot myelin water fraction (Deoni et al. 2008; our primary goal is to demonstrate a converging myelin Deoni and Kolind 2015), macromolecular tissue volume pattern of the corpus callosum using T1w/T2w MRI and (Mezer et al. 2013) and magnetization transfer (MT) meth- ex vivo post-mortem histology. Considering that myelin ods (Hagiwara et al. 2018; Henkelman et al. 2001; Stanisz and iron distributions overlap significantly in several brain et al. 1999). regions (Ogg and Steen 1998; Fukunaga et al. 2010), addi- Myelin is present in the cortex, but it is most abundant tional QSM MRI data were acquired on healthy subjects to in the cerebral white matter (Shafee et al. 2015), and the test whether T1w/T2w callosal maps are, instead, modu- ratio of T1-weighted/T2-weighted (T1w/T2w) MRI images lated by, or might reflect, iron density. Finally, a recently has been suggested as a reliable and straightforward in vivo released MWF atlas of the healthy human brain based on method to map myelin in the cerebral cortex (Glasser and multi-echo T2 MRI (Liu et al. 2019) has allowed a direct Van Essen, 2011). Some studies have even applied T1w/T2w comparison between the results obtained with T1w/T2w to white matter tracts in the neonatal (Lee et al. 2015; Chen and an alternative in vivo MRI myelin mapping technique. et al. 2017) and adult brain (Colmenares et al. 2021). How- ever, upon comparison with other myelin mapping imaging modalities, including Myelin Water Fraction (MWF), T1w/ Materials and methods T2w ‘may be not an optimal index of subcortical myelin content’ (Arshad et al. 2017), and its use as a myelin marker In vivo MRI data has been criticized (Uddin et al. 2018, 2019). Currently, there is still no histological evidence of the validity of T1w/ Myelin quantification: T1w/T2w maps T2w as a myelin mapping method when applied to white matter at the subcortical level. T1w/T2w structural datasets were downloaded from two This study, therefore, aims to clarify whether in  vivo distinct databases of the Human Connectome Project (HCP, T1w/T2w ratio is a method for mapping myelin density in http:// www. human conne ctome. org): HCP (57 healthy sub- white matter with (i) histological and (ii) imaging evidence. jects, aged 21–35) and HCP-Lifespan (10 healthy subjects, Specifically, we looked at in vivo T1w/T2w myelin mapping age 45–75). The reason behind adding ten more subjects of the corpus callosum from 67 healthy subjects from the from the latter dataset was to exclude any effect of age on Human Connectome Project (HCP) and we acquired ex vivo T1w/T2w. HCP data was already preprocessed according myelin staining from four post-mortem callosal samples. to the HCP preprocessing pipeline (described in Van Essen Additional quantitative susceptibility mapping (QSM) (Li et al. 2013), and made available also with the correspond- et  al. 2015) and MWF data (MacKay et al. 1994; Zhang ing transformation fields to MNI space. HCP-Lifespan data et al. 2015) were included to complement the MRI analysis. were processed by the authors following the same process- We focused on the corpus callosum because it repre- ing pipeline as available at https:// github. com/ Washi ngton- sents, among all the brain pathways, not only the largest Unive rsity/ Pipel ines. 1 3 Brain Structure and Function Briefly, this preprocessing pipeline involves three distinct approach is currently one of the standard methods to quan- pipelines: (1) PreFreeSurfer pipeline, to generate gradient tify myelin water fraction using MRI, and it is expected to distortion corrected, bias field (B1 inhomogeneities) cor - correlate with myelin density highly. In this study, we used rected T1w and T2w images, followed by brain extraction this atlas to compare MWF values with T1w/T2w maps and and registration to standard space; (2) FreeSurfer pipeline, histology in the callosal regions. to segment the structural volumes according to a specific parcellation, reconstruct white and pial cortical surfaces, Ex vivo myelin quantification: histological staining and apply surface registration to FreeSurfer surface atlas; and, finally, (3) the PostFreeSurfer pipeline, which produces Brain samples the final NIFTI files (including T1w/T2w maps) and GIFTI surface files registered to a surface atlas (Conte69), before We obtained four post-mortem brains at autopsy (with the downsampling and conversion to native space. After these prior informed consent of the person's relatives) from sub- steps, average surface cortical myelin maps were created jects without neurological or psychiatric diseases. To test using HCP Connectome Workbench (as described in Glasser whether we could replicate ex vivo the identical T1w/T2w and Van Essen 2011; Glasser et al. 2013). Similarly, average distribution pattern in the corpus callosum across different T1w/T2w MNI volumes for both HCP-Lifespan and HCP ages, three brains (male, 68 years; female, 73 years; male, were obtained using the computed MNI transformations and 82 years) from the Max Planck Institute for Human Cogni- FSL software package (fsl.fmrib.ox.ac.uk/fsl/fslwiki, Jen- tive and Brain Sciences were collected in Leipzig, Germany. kinson et al. 2012). One additional younger female brain (12 years) was obtained from King’s College Hospital, Department of Clinical Neu- Quantitative susceptibility mapping for iron content ropathology, London, United Kingdom. quantification All samples were formalin-fixed at 4% for more than 4 weeks before manually dissecting the corpus callosum. Data from 12 healthy subjects (70 ± 5 years, 6 female) were The three older brain samples were cryoprotected in a 30% acquired on a 3 T MR Biograph Siemens scanner using a sucrose solution and sectioned with a freezing microtome 12-channel coil. The protocol consisted in a spoiled gradi- (SM2000 R, Leica; Hyrax KS 34, Zeiss) in a sagittal plane ent-echo sequence with flow compensation and the follow - at 25 μm. A similar procedure was followed for the young ing parameters: TE = 19.6 ms, TR = 30 ms, flip angle = 12°, female brain where the sample was first formalin-fixed par - voxel size = 0.5 × 0.5 × 1  mm , matrix size = 478 × 378, 160 affin-embedded (FFPE) and then sectioned with a conven- slices. The Quantitative Susceptibility Maps (QSM) were tional microtome at 7 μm. obtained from phase images using the method described by Li et al. 2015. This approach includes the following steps: Myelin staining: luxol fast blue histology a sparse linear equation and least-squares algorithm-based method to derive an initial estimation of magnetic suscep- The sections were mounted on coated slides and stained for tibility; a fast-quantitative susceptibility mapping method myelin sheaths with an optimised Luxol Fast Blue (LFB) to estimate the susceptibility boundaries and an iterative protocol (Mulisch and Welsch 2015). The deparaffinised approach to estimate the susceptibility artifact from ill- sections were rinsed in distilled water for 5 min, and in 70, conditioned k-space regions only. The procedure generates 85, and 96% alcohol for 3 min each. They were incubated an unbiased estimate of tissue susceptibility with negligible overnight at a temperature of 60 °C in a 0,1% LFB solu- streaking artifacts. QSM maps of each subject were spatially tion (100 ml 96% alcohol, 0,1 g LFB (SERVA) and 0,5 ml normalised to MNI template by applying T1-driven spatial 10% acetic acid), followed by 96% alcohol for 3 min, and non-rigid transformation, estimated with SPM12; QSM distilled water for 5 min. They were subsequently rinsed in maps were voxel-wise averaged across all subjects to obtain 0.05% lithium carbonate for 5 seconds, differentiated two a QSM average map. times in 70% alcohol (30 s each time, with short movements to rinse out the excess colour), and rinsed twice in distilled Multi‑echo T2 myelin water fraction atlas water for 5 minutes. The sections were then dehydrated in graded alcohols, rinsed twice in Toluol and coverslipped An atlas of the average MWF distribution in the adult with Entellan (Merck). human brain has been made available at https://sour cef orge. net/ proje cts/ myelin- water- atlas/ (Liu et al. 2019). In brief, LFB optical density analysis this atlas was made using a multi-echo T2 approach, as described by MacKay et al. 1994, Zhang et al. 2015, using We digitized the sections of the three post-mortem older a 3D GRASE pulse sequence (Prasloski et al. 2012). This callosal samples with a digital camera (D90, Nikon), and 1 3 Brain Structure and Function the younger one with a photomicroscope with a motor- Statistical analysis ized stage (Axio Imager M1, Zeiss; 4 × magnification). We estimated myelin density using an LFB Optical Den- To quantitatively compare MRI maps with histological data, sity analysis approach (Beckmann et al. 2018) with the the same 92 ROIs sampling procedure was applied to the Nikon NIS-Elements Advanced Research 4.12 software. average T1w/T2w and MWF maps. Each map was sampled All acquisitions were made with the same light calibra- on two sagittal slices at + 4 and − 4 mm from the midsagit- tion. Three-channel RGB callosal images were converted tal section, and values were averaged. Histological LFB-OD to a single 256-grey scale. The 0–256 intensity range was data were averaged across the three adult callosal samples. rescaled from 1 to 7 and converted into a colour-coded Pearson correlations were performed to compare 92 ROIs map, where purple indicates the lowest level of myelin LFB-OD vs T1w/T2w, LFB-OD vs MWF and finally T1w/ density and dark red the highest level of myelin density T2w vs MWF. (Deshmukh et al. 2013). Results Definition of the callosal regions of interest (ROIs) Figure 1A illustrates the cortical distribution of the T1w/ A sampling over 92 circular regions of interest (ROIs) was T2w ratio in the 57 subjects from the HCP dataset. This performed following a template of the midsagittal section result coherently replicates previous findings from Glasser of the corpus callosum. As shown in the results section, and Van Essen (2011): high T1w/T2w ratio values can be the template is organised according to the anatomical and found in regions with high cortical myelination, such as geometrical subdivision proposed by Witelson 1989. It the primary motor-sensory and visual cortex. Figure  1B subdivides the corpus callosum into six macro-regions represents the volumetric distribution of T1w/T2w in the along the antero-posterior axis: rostrum/genu, rostral white matter for both the HCP and HCP-Life Span datasets. body, anterior midbody, posterior midbody, isthmus, and The highest T1w/T2w intensities is not in the white matter, splenium. The template is specifically designed based on but, instead, in grey matter regions such as the red nucleus, Witelson’s geometrical subdivision to take into account substantia nigra, globus pallidus and partially the dentate individual variability in callosal size among the subjects nucleus. This result is consistent in both the older and and allow comparisons among them (Witelson 1989). younger cohort. In the white matter of the corpus callosum, Fig. 1 T1w/T2w myelin maps at cortical and callosal level. A T1w/ (age 45–75, bottom). The highest intensity signals are located in the T2w myelin maps at the cortical level of 57 healthy participants red nuclei (RN) and globus pallidus (GP), while moderate and low (aged 21–35) from the Human Connectome Project. T1w/T2w corti- intensities are in the dentate nucleus (DN) and claustrum (CL). Right: cal maps show a gradient of T1w/T2w values from the primary motor averaged T1w/T2w map of the corpus callosum of the same subjects. (M1) and sensory areas (V1, S1, A1), which have higher values, to The colour-code map shows a heterogeneous distribution of T1w/ associative areas. This pattern is consistent with previous reports T2w values in the different callosal subregions, with the highest T1w/ (Nieuwenhuys 2013; Van Essen and Glasser 2014; Nieuwenhuys T2w values in the rostrum/genu and slightly lower in the isthmus/ et  al. 2015). Colour scale: purple, low T1w/T2w values; red, high splenium. Colour scale: purple, low T1w/T2w values; red, high T1w/ T1w/T2w values. B Left: averaged T1w/T2w maps of the whole brain T2w values of healthy participants from HCP (age 21–35, top) and HCP-Lifespan 1 3 Brain Structure and Function T1w/T2w maps reveal a heterogeneous distribution of values across samples of 0.62. An identical pattern, characterised along the callosal subregions, with higher T1w/T2w values by higher myelin content in the medium/posterior part of the in the rostrum/genu and isthmus/splenium compared to the corpus callosum, was also detected in the additional younger posterior midbody and the inferior splenium (Fig. 1B, right). brain sample (female, age: 12 years), which was dissected This pattern was consistently observed in both the HCP and and stained in a different laboratory (Fig.  3). HCP-Life Span datasets, thus suggesting that this intensity To further examine whether T1w/T2w maps are mod- heterogeneity is stable and not driven by age effects. ulated by iron content, we investigated the QSM map To test whether we could replicate ex vivo the identi- derived from an average of 12 healthy subjects (Fig.  4, cal T1w/T2w distribution pattern in the corpus callosum, top). As expected, regions with high QSM intensities we analysed three human post-mortem samples (ages: 68, are the red nucleus, substantia nigra and globus pallidus. 73, 82 years) stained with LFB. LFB staining can be seen These findings align well with what we found in the T1w/ in the histological slices (Fig. 2, centre) and with a 92 cir- T2w maps. However, regions such as the dentate nucleus cular ROIs sampling approach defined along the midsagit- tal sections of the entire corpus callosum (Fig.  2, right). The histological myelin distribution in the corpus callosum derived from LFB-optical density analysis is different from the T1w/T2w pattern. In fact, in all three LFB callosal sam- ples, myelin density in the anterior part (mainly rostrum/ genu) is low and higher in the posterior part (mainly isth- mus/splenium). This staining pattern is reproducible across a series of five consecutive sections for each case (see Sup- plementary Figs. 1–3). Indeed, the observed variance across the callosal regions was lower than the variance within the Fig. 3 Myelin mapping of the corpus callosum: LFB-stained midsag- ittal section of the corpus callosum of a 12-year-old female. Insets: different samples. For each of the 92 callosal data points, we LFB myelin staining at higher magnification shows lower intensity observed a normalised LFB intensity range extending from in the anterior corpus callosum (rostrum/genu, left inset) and higher values of 1 to 7, with an average standard deviation across intensity in the posterior part (isthmus, right inset) all regions of 1.47 but only an average standard deviation Fig. 2 Histology and ROI sampling. Left, top: one of the four post- subjects 1, 2 and 3. Right: ROI-based, colour-coded maps of the mye- mortem brain samples with the corpus callosum in situ. Left, bottom: lin distribution of subjects 1, 2 and 3 along the midsagittal section of template of the corpus callosum showing 92 circular regions of inter- the corpus callosum assessed ex vivo via histological myelin staining. est (ROIs) based on the anatomical subdivision proposed by Witelson The histological myelin distribution pattern of the corpus callosum is (1989). Values were calculated inside the circular ROIs in the centre different from the T1w/T2w pattern. All three LFB callosal samples of each of the 92 squares of the callosal template and then extended have low myelin density in the anterior part (mainly rostrum/genu) to the square surrounding the circle. Centre: Luxol Fast Blue (LFB)- and higher myelin density in the posterior part (mainly isthmus/sple- stained midsagittal sections  (25  µm thick) of the corpora callosa of nium). Colour scale: purple, low values; red, high values 1 3 Brain Structure and Function Fig. 4 QSM and MWF cortical and myelin mapping. Left, top: aver- sity in the dentate nucleus (DN) and claustrum (CL) is not matched in age quantitative susceptibility maps (QSM) illustrating iron distri- the T1w/T2w maps represented in Fig. 1. Left, bottom: Myelin Water bution in the whole brain in 12 healthy subjects (70 ± 5-year-old, 6 Fraction (MWF) maps from (Liu et  al. 2019) averaged MWF atlas. females). Higher intensities are localised in the red nuclei (RN)  and Right: averaged QSM and MWF maps of the corpus callosum show globus pallidus (GP) as in the T1w/T2w maps. The high QSM inten- different intensity patterns from anterior to posterior regions and claustrum have a high QSM intensity, but only moder- To quantitatively evaluate the correlation between T1w/ ate or low T1w/T2w values. At callosal level, QSM val- T2w, MWF and histology, we sampled the two MRI maps ues are high in the inferior part of the genu, rostrum and using the same 92-ROI sampling scheme used for histol- inferior splenium, where T1w/T2w values are also high. ogy. Figure 5 shows the correlation data: T1w/T2w does not However, the upper parts of the genu and splenium have positively correlate with LFB-Optical Density, but, instead, a lower mean magnetic susceptibility, whereas the T1w/ reveals a weak to moderate yet significant negative correla- T2w intensity is higher. A discrepancy is also evident for tion (r = 0.21, p < 0.001, Fig. 5A). On the contrary, MWF the anterior midbody (QSM values very low, T1w/T2w is strongly and positively correlated with LFB (r = 0.52, medium-to-high), whereas the isthmus reveals relatively p < 0.001, Fig. 5B), suggesting a good agreement between low intensities in both QSM and T1w/T2w maps. the two modalities. Finally, T1w/T2w and MWF maps are Compared to the MWF atlas, T1w/T2w maps suggest similarly weakly negatively correlated (r = 0.13, p < 0.001, a similar, increased intensity difference between white Fig. 5C). matter regions and cortical grey matter (Fig. 4, bottom). However, another pattern emerges when looking at differ - ences within white matter regions. The internal capsule Discussion and the region corresponding to the cortico-spinal tract display a high content of MWF, whereas T1w/T2w has an In this study, we applied multiple in vivo MRI and histo- opposite, darker contrast compared to more lateral white logical methods to clarify whether T1w/T2w is a method matter regions. By focusing on the body of the corpus cal- to map myelin in the white matter. Our results showed a losum, the MWF atlas is characterised by relatively high discrepancy between T1w/T2w maps and the alternative myelination in the anterior genu, but lower intensities in imaging modalities we applied. While histological studies the anterior body. Higher MWF intensities are evident have confirmed that cortical regions with high T1w/T2w from the posterior midbody and splenium, thus reproduc- signal co-localise with high myelin levels (Glasser et al. ing a pattern similar to the histological staining intensity. 2014), this contrast does not correlate with alternative 1 3 Brain Structure and Function Fig. 5 To quantitatively evaluate the correlation between T1w/T2w, significant negative correlation (r = 0.21 p < 0.001). B MWF pre- MWF and histology, MRI maps were sampled using the same 92-ROI sents a strong and significant positive correlation with LFB (r = 0.52, sampling scheme used for histology and correlated with the cor- p < 0.001), suggesting a good agreement between the two modalities. responding average LFB-OD values. Data is displayed on arbitrary C T1w/T2w and MWF maps do not correlate positively, as they are units (from 1 to 7) for all methods. A T1w/T2w does not positively characterised by a weak negative correlation (r = 0.13, p < 0.001) correlate with the LFB-Optical Density analysis but, instead, has a methods to quantify myelin in white matter regions. Myelin and iron distribution Overall, the T1w/T2w ratio reveals increased values in white matter regions compared to cortical grey matter, The discrepancy we found at the callosal level suggests that but, within white matter, its intensity variability does other biological factors should be considered when these not follow the same pattern as illustrated by the other maps are applied to the white matter. We adopted LFB as methods. Quite striking is the relatively low T1w/T2w myelin staining, which is not only one of the most used intensity in the regions corresponding to the corticospinal staining methods globally and the reference staining tool tract and internal capsule compared to the other white used in routine diagnostics (Kluver and Barrera 1953; Laz- matter regions with higher intensity. This pattern is not zarini 2003), but it has also been widely used in previous found in the MWF atlas, where the internal capsule and papers to compare myelin mapping MRI methods quantita- the corticospinal have a higher myelin content than white tively and to quantify changes in diseases or animal models matter regions. When mapped against the callosal histol- (Laule et al. 2006; Khodanovich et al. 2017; Beckmann et al. ogy, T1w/T2w seems again to show a different pattern 2018; Wood et al. 2016). compared to LFB-OD analysis. In T1w/T2w, the genu Luxol Fast Blue stain ‘highlights the blue myelinated is the region with the highest ‘myelination’, followed by axons of neurons in the white matter of the nervous system the superior portion of the splenium. On the other hand, and the small dense round nuclei of oligodendrocytes that while indicating an increase in myelination in the most produce this myelin sheath’ (Lindberg and Lamps 2018). anterior part of the genu, LFB permits observing much Myelin and iron distributions co-localise significantly in stronger myelination in the posterior regions of the corpus many regions (Ogg and Steen 1998; Fukunaga et al. 2010), callosum, which is consistent across all the histological especially in the visual cortex and in the motor/somatosen- samples and the subjects, irrespective of age and gen- sory cortex (Stüber et al. 2014), and prior MRI investiga- der. This is further confirmed by the correlation analysis tions of brain iron have been published (Gelman et al. 1999; based on the 92-ROI sampling, where T1w/T2w does not Haacke et al. 2005). Therefore, a plausible alternative expla- positively correlate with the LFB nor the MWF measure- nation is that T1w/T2w maps are modulated by, or might ments. On the contrary, MWF and LFB have a strong reflect, differences in iron density. Our results support this positive correlation (r = 0.72, r = 0.52, p < 0.001), thus hypothesis, as very high T1w/T2w intensities were recorded supporting the idea that both modalities indeed capture in regions that are known for their high iron content, includ- myelin density. These observations suggest that the T1w/ ing the globus pallidus, substantia nigra and red nucleus T2w contrast may not represent a suitable method for (Rouault 2013; Piñero and Connor 2000, but see also Koep- mapping the myelin in the corpus callosum and, arguably, pen 1995). These high-intensity regions are evident in in the white matter in general. both the HCP-Lifespan dataset with older subjects and the 1 3 Brain Structure and Function younger group from HCP, a piece of evidence that seems to is needed to clarify the biological meaning and elucidate the rule out any age effect. T1w/T2w signal's complexity. Our QSM results confirm this overlap with iron at the Other MRI methods have been proposed to quantify mye- level of these subcortical regions. However, this overlap is lin mapping, such as multi-echo-T2 (MacKay et al. 1994; not present for other subcortical regions, such as the den- Zhang et al. 2015), mcDespot (Deoni et al. 2008; Deoni tate nucleus and claustrum. Similarly, relatively high signal and Kolind 2015), quantitative T1 maps (Geyer et al. 2011; intensity is evident for both T1w/T2w and QSM maps in the Geyer and Turner 2013; Bock et al. 2013), and macromo- inferior genu and the inferior splenium, but not in the supe- lecular tissue volume (Mezer et al. 2013). All these methods rior genu, superior splenium and anterior midbody. In light attempt to provide a quantitative measure of myelin, and of this, it seems appropriate to conclude that the iron content their histological validation is currently ongoing (Lazari may modulate the T1w/T2w ratio only in some subcortical and Lipp 2021; Laule et al. 2006; Laule 2008; Dula et al. nuclei, and possibly some subregions of the corpus callo- 2010; Fatemi et al. 2011; Khodanovich et al. 2017). To date, sum, but it does not fully explain the origin of the T1w/T2w whether myelin mapping biomarkers can give absolute and contrast. The lack of published histologically defined iron quantitative indications of myelin density is still an open maps of the corpus callosum does not allow us to directly question. Nevertheless, T1w/T2w maps cannot be fully compare myelin vs. iron distribution to test further hypoth- considered quantitative because they heavily depend on the eses beyond our QSM results. actual scanning parameters (i.e., multiple T1w and T2w con- MWF maps correlate well with histology and seem trasts can be obtained on the same subject by changing the good myelin maps. However, various structures definitively MR acquisition parameters), and this effectively changes the known not to contain myelin also show up with high MWF, final contrast and their reproducibility. Therefore, T1w/T2w including the dura (especially the falx in the publicly avail- maps should be assumed to provide qualitative information, able atlas), the cerebral arteries and veins, and the extraocu- and particular care should be taken when comparing results lar muscles. Moreover, MWF images seem strongly affected across studies or scanner manufacturers. by iron (e.g., in the globus pallidus), similar to T1w/T2w, NODDI, T2w and quantitative T2. Overall, it can be intrigu- Limitations of the study ing to speculate on the potential differences in the iron pres- ence (ferritin, within microglia, etc.) and if and how that This study has some limitations. Our group of older par- may impact QSM and T1/T2 measures differently, although ticipants encompassed ten subjects. However, by comparing this is clearly outside the scope of the paper and will be T1w/T2w maps in a younger cohort of 57 HCP participants, clarified by future works. we obtained identical results, and we concluded that T1w/ T2w findings are not age-dependent. Another limitation is Myelin and axonal diameter that our report is based on an indirect comparison between in vivo and ex vivo maps acquired from different subjects. An alternative hypothesis is that the T1w/T2w signal might Ideally, MRI and histology should have been acquired from also be modulated by the complex distribution of different the same subjects. This is certainly feasible with animal populations of fibres with varying axonal diameters. His- models or by taking advantage of human post-mortem imag- tological maps of the axonal diameter distribution along ing, albeit for the latter the correspondence between in vivo the corpus callosum have been published by Aboitiz et al. T1w/T2w and post-mortem T1w/T2w still needs to be veri- (1992), and their pattern of distribution strikingly overlaps fied, and this would have introduced a further source of with our T1w/T2w maps. In particular, the higher T1w/ variability. Similarly, the QSM map and the MWF atlas are T2w signal in the anterior part of the corpus callosum cor- derived from different groups of individuals. However, as we responds to a region containing small-diameter axons. In detected the same T1w/T2w pattern for different cohorts and contrast, the central segment of the corpus callosum with the same pattern in each histology sample, we expect that a lower T1w/T2w signal corresponds to a region with pre- what is measured in this study is overall robust and beyond dominantly large-diameter axons. Furthermore, the mixed interindividual variability. Nevertheless, an approach that low–high-low intensity pattern in the posterior corpus cal- could be applied in future studies would be to collect QSM, losum parallels the large–small–large pattern of the axonal multi-echo T2-MWF and T1w/T2w on the same group of diameter distribution (Aboitiz et al. 1992). subjects. Overall, at this point, we cannot exclude that the T1w/ We adopted LFB as myelin staining for the reasons T2w signal is a rather aspecific contrast modulated by mul- explained above, but we are aware that other histological tiple biological factors. In fact, in addition to myelin, iron methods are available, including staining for myelin basic and fibre density, other glial cells, elements of extracellular protein (MBP) and myelin proteolipid protein (PLP). Moreo- space and vasculature may play a role. Additional research ver, a quantification of the callosal myelin density has been 1 3 Brain Structure and Function performed in the macaque brain using high-resolution Elec- resolution, which, in turn, might lead to significant partial tron Microscopy (EM) data (Stikov et al. 2015a and b). How- voluming and rather long acquisition time. MWF maps ever, while EM can provide accurate and precise quantifica- correlate well with histology. However, considering the tion of myelin, only small sample regions can be selected. non-myelinated structures displaying high MWF values, If samples and tissues are not homogenous, this may lead it is plausible that these are not entirely specific for myelin to notable sampling bias. Considering that our human sam- either. The quest for the most suitable MRI measure for ples were fixed by immersion, compared to perfusion fixa- quantifying myelin is still open (Mancini et al. 2020), and tion used in animal models, this may have introduced more future studies can potentially extend this finding by using local inhomogeneities, making the final interpretation of the also MT-based methods (Hagiwara et al. 2018; Henkelman results more difficult. et al. 2001). On a more general note, methodological and In this paper, we used the anatomical and geometrical performance variability, inconsistent reporting, publica- subdivision of the corpus callosum proposed by Witelson tion bias and limited validation data available are report- (1989). We are aware that more refined parcellation sub- edly some of the factors limiting the comparison between divisions have been proposed, such as those by Hofer and MRI-based myelin markers (Lazari and Lipp 2021; van der Frahm (2006) and Caminiti et al. (2013), to study the precise Weijden et al. 2021). topography of callosal projections. However, the choice of the Witelson subdivision was driven by the need to (i) com- pare our results with those obtained in the reference work by Aboitiz et al. (1992), (ii) define a reproducible geometrical Conclusions constraint for the 92 regions of interest used to quantify mye- lin density across the whole corpus callosum, (iii) to provide Developing a reliable in vivo method for myelin mapping comparable anatomical information with previous studies is crucial to understanding myelination in healthy and (Aboitiz et al. 1992; Zaidel and Iacoboni 2003; Catani and pathological conditions. This paper represents a prelimi- Thiebaut de Schotten 2012). nary attempt to clarify the biological underpinnings of the Finally, the handling and processing of post-mortem T1w/T2w signal. The results highlight important limita- brains involved many steps that could potentially introduce tions in interpreting T1w/T2w as a valid myelin contrast biases. We cannot exclude that some of the inter-individual in white matter. The discrepancy between T1w/T2w MRI variabilities might be related to differences in slice location maps, QSM, MWF and histological myelin maps sug- or inhomogeneity along the slice due to cutting artefacts or gests caution in using T1w/T2w as a white matter map- vascularization. However, these are unlikely to impact our ping method. Future studies are needed to clarify the exact findings as the same pattern of myelin density was preserved biological meaning and the origin of the T1w/T2w signal and commonly present in all the consecutive stained callosal in the white matter. slices, beyond inter-individual differences and across differ - Supplementary Information The online version contains supplemen- ent laboratories (Supplementary Figs. 1–3). tary material available at https://doi. or g/10. 1007/ s00429- 022- 02600-z . Although the paper focuses on ‘putting to the test’ the T1w/T2w method as a myelin mapping tool, it is worth stat- Funding Stefano Sandrone was funded by the National Institute for Health Research (NIHR) Biomedical Research Centre at South London ing that all the imaging methods have strengths and weak- and Maudsley NHS Foundation Trust and King’s College London. nesses.  For example, T1w/T2w maps’ strengths include Marco Catani is the recipient of a Wellcome Trust Investigator Award ease of acquisition on 3 T scanners, high spatial resolution, (103759/Z/14/Z). The London Neurodegenerative Diseases Brain Bank receives funding from the MRC and as part of the Brains for Demen- high contrast-to-noise ratio (CNR), short acquisition times tia Research programme, jointly funded by Alzheimer’s Research UK (Glasser and Van Essen 2011). Furthermore, T1w/T2w and Alzheimer’s Society. Michel Thiebaut de Schotten is funded by images are useful for other kinds of processing, including the European Research Council (ERC) under the European Union’s subcortical and cortical segmentation or surface generation Horizon 2020 research and innovation programme (Grant agreement No. 818521). Flavio Dell’Acqua is funded by the Sackler Institute for (Glasser et al. 2014). At the same time, these are not quan- Translational Neurodevelopment. Michel Thiebaut de Schotten would titative, and there may be a second order anti-correlation like to thank the University of Bordeaux−s IdEx “Investments for the with other white matter measures related to the predominant Future” program RRI “IMPACT,” which received financial support axonal diameter in the white matter. from the French government. We thank the staff of the Clinical Neuro- pathology Department at King’s College Hospital for diagnostic and Moreover, these are typically harder to acquire at the technical assistance, and we also thank the donors and their families. same acquisition resolution, field strength, acquisition time, and CNR as T1w/T2w. MWF images are closer Data availability This manuscript has no associated data. to the histological profile, as shown here. Yet the pub- licly available datasets seem to have a very low spatial 1 3 Brain Structure and Function A regenerative approach to the treatment of multiple sclerosis. Declarations Nature 502(7471):327–332 Dula AN, Gochberg DF, Valentine HL, Valentine WM, Does MD Conflict of interest Stefano Sandrone has no relevant financial or non- (2010) Multiexponential T2, magnetization transfer, and quan- financial interests to disclose. titative histology in white matter tracts of rat spinal cord. Magn Reson Med 63(4):902–909 Open Access This article is licensed under a Creative Commons Attri- Fatemi A, Wilson MA, Phillips AW, McMahon MT, Zhang J, Smith bution 4.0 International License, which permits use, sharing, adapta- SA, Arauz EJ, Falahati S, Gummadavelli A, Bodagala H, Mori S, tion, distribution and reproduction in any medium or format, as long Johnston MV (2011) In vivo magnetization transfer MRI shows as you give appropriate credit to the original author(s) and the source, dysmyelination in an ischemic mouse model of periventricular provide a link to the Creative Commons licence, and indicate if changes leukomalacia. J Cereb Blood Flow Metab 31(10):2009–2018 were made. The images or other third party material in this article are Fukunaga M, Li TQ, van Gelderen P, de Zwart JA, Shmueli K, included in the article's Creative Commons licence, unless indicated Yao B, Lee J, Maric D, Aronova MA, Zhang G, Leapman RD, otherwise in a credit line to the material. If material is not included in Schenck JF, Merkle H, Duyn JH (2010) Layer-specific variation the article's Creative Commons licence and your intended use is not of iron content in cerebral cortex as a source of MRI contrast. permitted by statutory regulation or exceeds the permitted use, you will Proc Natl Acad Sci U S A 107(8):3834–3839 need to obtain permission directly from the copyright holder. To view a Gelman N, Gorell JM, Barker PB, Savage RM, Spickler EM, Wind- copy of this licence, visit http://cr eativ ecommons. or g/licen ses/ b y/4.0/ . ham JP, Knight RA (1999) MR imaging of human brain at 3.0 T: preliminary report on transverse relaxation rates and relation to estimated iron content. Radiology 210(3):759–767 References Geyer S, Turner R (2013) Microstructural parcellation of the human cerebral cortex. Springer-Verlag, Berlin Geyer S, Weiss M, Reimann K, Lohmann G, Turner R (2011) Micro- Aboitiz F, Scheibel AB, Fisher RS, Zaidel E (1992) Fiber composi- structural parcellation of the human cerebral cortex-From Brod- tion of the human corpus callosum. Brain Res 598(1–2):143–153 mann’s post-mortem map to in vivo mapping with high-field Alberts B, Alberts B, Johnson A, Lewis J, Raff M, Roberts K, Walter P magnetic resonance imaging. Front Hum Neurosci 5:19 (2014) Molecular biology of the cell. Garland Science, New York Glasser MF, Van Essen DC (2011) Mapping human cortical Arshad M, Stanley JA, Raz N (2017) Test–retest reliability and con- areas in vivo based on myelin content as revealed by T1 and current validity of in vivo myelin content indices: myelin water T2-weighted MRI. J Neurosci 31(32):11597–11616 fraction and calibrated T1w/T2w image ratio. Hum Brain Mapp Glasser MF, Goyal MS, Preuss TM, Raichle ME, Van Essen DC 38(4):1780–1790 (2014) Trends and properties of human cerebral cortex: correla- Beckmann N, Giorgetti E, Neuhaus A et al (2018) Brain region-specific tions with cortical myelin content. Neuroimage 93(2):165–175 enhancement of remyelination and prevention of demyelination Glasser MF, Sotiropoulos SN, Wilson JA, Coalson TS, Fischl B, by the CSF1R kinase inhibitor BLZ945. Acta Neuropathol Com- Andersson JL, Xu J, Jbabdi S, Webster M, Polimeni JR, Van mun 6(1):9 Essen DC, Jenkinson M, WU-Minn HCP (2013) The minimal Bock NA, Hashim E, Janik R, Konyer NB, Weiss M, Stanisz GJ, Turner preprocessing pipelines for the human connectome project. R, Geyer S (2013) Optimizing T1-weighted imaging of cortical Neuroimage 15(80):105–124 myelin content at 3.0 T. Neuroimage 65:1–12 Haacke EM, Cheng NY, House MJ, Liu Q, Neelavalli J, Ogg RJ, Caminiti R, Ghaziri H, Galuske R, Hof PR, Innocenti GM (2009) Khan A, Ayaz M, Kirsch W, Obenaus A (2005) Imaging iron Evolution amplified processing with temporally dispersed slow stores in the brain using magnetic resonance imaging. Magn neuronal connectivity in primates. Proc Natl Acad Sci USA Reson Imaging 23(1):1–25 106(46):19551–19556 Hagiwara A, Hori M, Kamagata K, Warntjes M, Matsuyoshi D, Caminiti R, Carducci F, Piervincenzi C, Battaglia-Mayer A, Confa- Nakazawa M, Ueda R, Andica C, Koshino S, Maekawa T, Irie lone G, Visco-Comandini F, Pantano P, Innocenti GM (2013) R, Takamura T, Kumamaru KK, Abe O, Aoki S (2018) Myelin Diameter, length, speed, and conduction delay of callosal axons measurement: comparison between simultaneous tissue relax- in macaque monkeys and humans: comparing data from histol- ometry, magnetization transfer saturation index, and T1w/T2w ogy and magnetic resonance imaging diffusion tractography. J ratio methods. Sci Rep 8(1):1–12 Neurosci 33(36):14501–14511 Henkelman RM, Stanisz GJ, Graham SJ (2001) Magnetization trans- Catani M, Thiebaut de Schotten M (2012) Atlas of human brain con- fer in MRI: a review. NMR Biomed 14(2):57–64 nections. Oxford University Press, Oxford Hofer S, Frahm J (2006) Topography of the human corpus callosum Chen H, Budin F, Noel J, Prieto JC, Gilmore J, Rasmussen J, Wad- revisited-comprehensive fiber tractography using diffusion ten- hwa PD, Entringer S, Buss C, Styner M (2017) White mat- sor magnetic resonance imaging. Neuroimage 32(3):989–994 ter fiber-based analysis of T1w/T2w ratio map. Med Imag Jenkinson M, Beckmann CF, Behrens TEJ, Woolrich MW, Smith SM 2017(10133):195–201 (2012) FSL. Neuroimage 62(2):782–790 Colmenares AM, Voss MW, Fanning J, Salerno EA, Gothe NP, Kandel ER, Schwartz JH, Jessell TM, Siegelbaum SA, Hudspeth AJ Thomas ML, Edward McAuley E, Kramer AF, Burzynska AZ (2012) Principles of neural science. McGraw-Hill Professional, (2021) White matter plasticity in healthy older adults: the effects New York of aerobic exercise. Neuroimage 239:118305 Khodanovich MY, Sorokina IV, Glazacheva VY et al (2017) Histologi- Deoni SC, Kolind SH (2015) Investigating the stability of mcDESPOT cal validation of fast macromolecular proton fraction mapping as a myelin water fraction values derived using a stochastic region quantitative myelin imaging method in the cuprizone demyelina- contraction approach. Magn Reson Med 73(1):161–169 tion model. Sci Rep 7:46686 Deoni SC, Rutt BK, Arun T, Pierpaoli C, Jones DK (2008) Gleaning Kluver H, Barrera E (1953) A method for the combined staining of multicomponent T1 and T2 information from steady-state imaging cells and fibers in the Nervous system. J Neuropath Exp Neurol data. Magn Reson Med 60(6):1372–1387 12:400–403 Deshmukh VA, Tardif V, Lyssiotis CA, Green CC, Kerman B, Kim Koeppen AH (1995) The history of iron in the brain. J Neurol Sci HJ, Padmanabhan K, Swoboda JG, Ahmad I, Kondo T, Gage FH, 134(Suppl):1–9 Theofilopoulos AN, Lawson BR, Schultz PG, Lairson LL (2013) 1 3 Brain Structure and Function Laule C (2008) Myelin water imaging of multiple sclerosis at 7 T: Rouault TA (2013) Iron metabolism in the CNS: implications for neu- correlations with histopathology. Neuroimage 40(4):1575–1580 rodegenerative diseases. Nat Rev Neurosci 14(8):551–564 Laule C, Leung E, Lis DK et al (2006) Myelin water imaging in multi- Schmahmann JD, Pandya DN (2006) Fiber pathways of the brain. ple sclerosis: quantitative correlations with histopathology. Mult Oxford University Press, New York Scler 12(6):747–753 Shafee R, Buckner RL, Fischl B (2015) Gray matter myelination of Lazari A, Lipp I (2021) Can MRI measure myelin? Systematic review, 1555 human brains using partial volume corrected MRI images. qualitative assessment, and meta-analysis of studies validating Neuroimage 105:473–485 microstructural imaging with myelin histology. Neuroimage Stanisz GJ, Kecojevic A, Bronskill MJ, Henkelman RM (1999) Char- 230:117744 acterizing white matter with magnetization transfer and T2. Mag Lazzarini RA (2003) Myelin biology and disorders. Elsevier Academic Res Med 42(6):1128–1136 Press, San Diego Stikov N, Campbell JS, Stroh T et al (2015a) Quantitative analysis Lee K, Cherel M, Budin F, Gilmore J, Consing KZ, Rasmussen J, of the myelin g-ratio from electron microscopy images of the Wadhwa PD, Entringer S, Glasser MF, Van Essen DC, Buss C, macaque corpus callosum. Data Brief 4:368–373 Styner M (2015) Early postnatal myelin content estimate of white Stikov N, Campbell JS, Stroh T et al (2015b) In vivo histology of the matter via T1w/T2w ratio. Med Imag 9417:484–490 myelin g-ratio with magnetic resonance imaging. Neuroimage Li W, Wang N, Yu F, Han H, Cao W, Romero R, Tantiwongkosi B, 118:397–405 Duong TQ, Liu C (2015) A method for estimating and removing Stüber C, Morawski M, Schäfer A, Labadie C, Wähnert M, Leuze streaking artifacts in quantitative susceptibility mapping. Neuro- C, Streicher M, Barapatre N, Reimann K, Geyer S, Spemann image 108:111–122 D, Turner R (2014) Myelin and iron concentration in the Lindberg MR, Lamps LW (2018) Diagnostic pathology: normal histol- human brain: a quantitative study of MRI contrast. Neuroimage ogy. Elsevier, Amsterdam 93(1):95–106 Liu H, Rubino C, Dvorak AV, Jarrett M, Ljungberg E, Vavasour IM, Uddin MN, Figley TD, Solar KG, Shatil AS, Figley CR (2019) Com- Lee LE, Kolind SH, MacMillan EL, Traboulsee A, Lang DJ, parisons between multi-component myelin water fraction, T1w/ Rauscher A, Li DKB, MacKay AL, Boyd LA, Kramer JLK, Laule T2w ratio, and diffusion tensor imaging measures in healthy C (2019) Myelin water atlas: a template for myelin distribution in human brain structures. Sci Rep 9(1):1–17 the brain. J Neuroimaging 29(6):699–706 Uddin MN, Figley TD, Marrie RA, Figley CR, CCOMS Study Group MacKay A, Whittall K, Adler J, Li D, Paty D, Graeb D (1994) In Vivo (2018) Can T1w/T2w ratio be used as a myelin-specific meas - visualization of myelin water in brain by magnetic resonance. ure in subcortical structures? Comparisons between FSE-based Magn Reson Med 31:673–677 T1w/T2w ratios, GRASE-based T1w/T2w ratios and multi-echo Mancini M, Karakuzu A, Cohen-Adad J, Cercignani M, Nichols TE, GRASE-based myelin water fractions. NMR Biomed 31(3):e3868 Stikov N (2020) An interactive meta-analysis of MRI biomarkers van der Knaap MS, Valk J (2011) Magnetic resonance of myelination of myelin. Elife 9:e61523 and myelin disorders. Springer, Berlin Mezer A, Yeatman JD, Stikov N, Kay KN, Cho NJ, Dougherty RF, van der Weijden CWJ, García DV, Borra RJH, Thurner P, Meilof JF, Perry ML, Parvizi J, le Hua H, Butts-Pauly K, Wandell BA van Laar PJ, Dierckx RAJO, Gutmann IW, de Vries EFJ (2021) (2013) Quantifying the local tissue volume and composition in Myelin quantification with MRI: a systematic review of accuracy individual brains with magnetic resonance imaging. Nat Med and reproducibility. Neuroimage 226:117561 19(12):1667–1672 Van Essen DC, Smith SM, Barch DM, Behrens TE, Yacoub E, Ugurbil Mulisch M, Welsch U (2015) Romeis mikroskopische technik. Springer K, WU-Minn HCP Consortium (2013) The WU-Minn Human Spektrum, Berlin Connectome project: an overview. Neuroimage 15(80):62–79 Nieuwenhuys R (2013) The myeloarchitectonic studies on the human Van Essen DC, Glasser MF (2014) In vivo architectonics: a cortico- cerebral cortex of the Vogt-Vogt school, and their significance for centric perspective. Neuroimage 93(2):157–164 the interpretation of functional neuroimaging data. Brain Struct Witelson SF (1989) Hand and sex differences in the isthmus and genu Funct 218(2):303–352 of the human corpus callosum. Postmortem Morphol Study Brain Nieuwenhuys R, Broere CA, Cerliani L (2015) A new myeloarchitec- 112(3):799–835 tonic map of the human neocortex based on data from the Vogt- Wood TC, Simmons C, Hurley SA et al (2016) Whole-brain ex-vivo Vogt school. Brain Struct Funct 220(5):2551–2573 quantitative MRI of the cuprizone mouse model. PeerJ 4:e2632 Nieuwenhuys R, Voogd J, van Huijzen C (2007) The human central Zaidel E, Iacoboni M (2003) The parallel brain. the cognitive neurosci- nervous system: a synopsis and atlas. Steinkopff, Amsterdam ence of the corpus callosum Cambridge. MIT University Press, Ogg RJ, Steen RG (1998) Age-related changes in brain T1 are corre- Cambridge lated with iron concentration. Magn Reson Med 40(5):749–753 Zhang J, Kolind SH, Laule C, MacKay AL (2015) Comparison of Piñero DJ, Connor JR (2000) Iron in the brain: an important contrib- myelin water fraction from multiecho T2 decay curve and steady- utor in normal and diseased states. Neuroscientist 6(6):435–453 state methods. Magn Reson Med 73(1):223–232 Prasloski T, Rauscher A, MacKay AL, Hodgson M, Vavasour IM, Laule C, Mädler B (2012) Rapid whole cerebrum myelin Publisher's Note Springer Nature remains neutral with regard to water imaging using a 3D GRASE sequence. Neuroimage jurisdictional claims in published maps and institutional affiliations. 63(1):533–539 Quarles RH, Macklin WB, Morell P (2006) Myelin formation, struc- ture and biochemistry. In: Siegel GJ, Albers RW, Brady ST, Price D (eds) Basic Neurochemistry: molecular, cellular and medical aspects. Academic Press Elsevier, New York 1 3 http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png Brain Structure and Function Springer Journals

Mapping myelin in white matter with T1-weighted/T2-weighted maps: discrepancy with histology and other myelin MRI measures

Download PDF
11 pages

Loading next page...
 
/lp/springer-journals/mapping-myelin-in-white-matter-with-t1-weighted-t2-weighted-maps-xqjTSVNg4h

References (71)

Publisher
Springer Journals
Copyright
Copyright © The Author(s) 2023
ISSN
1863-2653
eISSN
1863-2661
DOI
10.1007/s00429-022-02600-z
Publisher site
See Article on Publisher Site

Abstract

The ratio of T1-weighted/T2-weighted magnetic resonance images (T1w/T2w MRI) has been successfully applied at the cortical level since 2011 and is now one of the most used myelin mapping methods. However, no reports have explored the histological validity of T1w/T2w myelin mapping in white matter. Here we compare T1w/T2w with ex vivo postmortem histology and in vivo MRI methods, namely quantitative susceptibility mapping (QSM) and multi-echo T2 myelin water fraction (MWF) mapping techniques. We report a discrepancy between T1w/T2w myelin maps of the human corpus callo- sum and the histology and analyse the putative causes behind such discrepancy. T1w/T2w does not positively correlate with Luxol Fast Blue (LFB)-Optical Density but shows a weak to moderate, yet significant, negative correlation. On the contrary, MWF is strongly and positively correlated with LFB, whereas T1w/T2w and MWF maps are weakly negatively correlated. The discrepancy between T1w/T2w MRI maps, MWF and histological myelin maps suggests caution in using T1w/T2w as a white matter mapping method at the callosal level. While T1w/T2w imaging may correlate with myelin content at the cortical level, it is not a specific method to map myelin density in white matter. Keywords Corpus callosum · Myelin mapping · Neuroanatomy · Neuroimaging · Validation · Quantitative susceptibility mapping · Myelin water fraction Marco Catani and Flavio Dell’Acqua have equally contributed to the paper. * Stefano Sandrone Department of Neurophysics, Max Planck Institute sandrone.stefano@gmail.com for Human Cognitive and Brain Sciences, Leipzig, Germany * Flavio Dell’Acqua Paul Flechsig Institute of Brain Research, University flavio.dellaqua@kcl.ac.uk of Leipzig, Leipzig, Germany Department of Basic and Clinical Neuroscience, Institute Department of Brain Sciences, Faculty of Medicine, of Psychiatry, Psychology & Neuroscience, King’s College Imperial College London, London, UK London, London, UK NatBrainLab, Department of Forensic Clinical Neuropathology, King’s College Hospital NHS and Neurodevelopmental Sciences, Institute of Psychiatry, Foundation Trust, London, UK Psychology & Neuroscience, King’s College London, London, UK Department of Neuroimaging, Institute of Psychiatry, Psychology & Neuroscience, King’s College London, The Sackler Institute for Translational Neurodevelopment, London, UK Institute of Psychiatry, Psychology & Neuroscience, King’s College London, London, UK Developmental Imaging and Biophysics Section, UCL Great Ormond Street Institute of Child Health, London, UK IRCCS SYNLAB SDN S.p.A., Naples, Italy Institute of Biostructures and Bioimaging, National Research Brain Connectivity and Behaviour Laboratory, Sorbonne Council, Naples, Italy Universities, Paris, France Groupe d’Imagerie Neurofonctionnelle, Institut des Maladies Neurodégénératives-UMR 5293, CNRS, CEA University of Bordeaux, Bordeaux, France Vol.:(0123456789) 1 3 Brain Structure and Function white matter tract, but also the most thoroughly examined Introduction using in vivo and post-mortem techniques in animals and humans (Zaidel and Iacoboni 2003). A further advantage Half of the human brain is composed of white matter tis- is that an established geometric subdivision of the cor- sue, which primarily contains myelinated axons of various pus callosum is thought to reflect different cortical pro- diameters (Alberts et al. 2014; Quarles et al. 2006). Myelin jections (Witelson 1989). According to this subdivision, sheath surrounds axons and is essential for the efficient the rostrum and the genu project to the prefrontal cortex, conduction of action potential between neurons (Nieu- whereas the posterior body and the splenium project to the wenhuys et al. 2007; Kandel et al. 2012). Several human sensory-motor and visual areas, respectively (Zaidel and brain diseases are associated with white matter pathology Iacoboni 2003; Catani and Thiebaut de Schotten 2012). (Lazzarini 2003; Schmahmann and Pandya 2006; van der Microscopy studies of the corpus callosum demonstrated Knaap and Valk 2011; Catani and Thiebaut de Schotten a heterogeneous distribution of fibres along the antero- 2012). It is, therefore, crucial to have reliable and valid posterior axis (Aboitiz et al. 1992; Zaidel and Iacoboni methods to map brain myelination in the living brain. 2003; Caminiti et al. 2009, 2013). Unmyelinated fibres Advances in structural imaging include approaches make up 16% of the genu and under 5% of other callosal towards myelin mapping, such as the ratio of T1-weighted/ areas; moreover, the posterior midbody is the region with T2-weighted Magnetic Resonance Imaging (MRI) scans the highest density of large and highly myelinated callosal (Glasser and Van Essen 2011; Glasser et al. 2014), multi fibres (Aboitiz et al. 1992). This permits us to put forward echo-T2 (MacKay et al. 1994; Zhang et al. 2015), T1 maps specific hypotheses about the topographical distribution (Geyer et al. 2011; Geyer and Turner 2013; Bock et al. of the myelin density along the corpus callosum. Hence, 2013), mcDespot myelin water fraction (Deoni et al. 2008; our primary goal is to demonstrate a converging myelin Deoni and Kolind 2015), macromolecular tissue volume pattern of the corpus callosum using T1w/T2w MRI and (Mezer et al. 2013) and magnetization transfer (MT) meth- ex vivo post-mortem histology. Considering that myelin ods (Hagiwara et al. 2018; Henkelman et al. 2001; Stanisz and iron distributions overlap significantly in several brain et al. 1999). regions (Ogg and Steen 1998; Fukunaga et al. 2010), addi- Myelin is present in the cortex, but it is most abundant tional QSM MRI data were acquired on healthy subjects to in the cerebral white matter (Shafee et al. 2015), and the test whether T1w/T2w callosal maps are, instead, modu- ratio of T1-weighted/T2-weighted (T1w/T2w) MRI images lated by, or might reflect, iron density. Finally, a recently has been suggested as a reliable and straightforward in vivo released MWF atlas of the healthy human brain based on method to map myelin in the cerebral cortex (Glasser and multi-echo T2 MRI (Liu et al. 2019) has allowed a direct Van Essen, 2011). Some studies have even applied T1w/T2w comparison between the results obtained with T1w/T2w to white matter tracts in the neonatal (Lee et al. 2015; Chen and an alternative in vivo MRI myelin mapping technique. et al. 2017) and adult brain (Colmenares et al. 2021). How- ever, upon comparison with other myelin mapping imaging modalities, including Myelin Water Fraction (MWF), T1w/ Materials and methods T2w ‘may be not an optimal index of subcortical myelin content’ (Arshad et al. 2017), and its use as a myelin marker In vivo MRI data has been criticized (Uddin et al. 2018, 2019). Currently, there is still no histological evidence of the validity of T1w/ Myelin quantification: T1w/T2w maps T2w as a myelin mapping method when applied to white matter at the subcortical level. T1w/T2w structural datasets were downloaded from two This study, therefore, aims to clarify whether in  vivo distinct databases of the Human Connectome Project (HCP, T1w/T2w ratio is a method for mapping myelin density in http:// www. human conne ctome. org): HCP (57 healthy sub- white matter with (i) histological and (ii) imaging evidence. jects, aged 21–35) and HCP-Lifespan (10 healthy subjects, Specifically, we looked at in vivo T1w/T2w myelin mapping age 45–75). The reason behind adding ten more subjects of the corpus callosum from 67 healthy subjects from the from the latter dataset was to exclude any effect of age on Human Connectome Project (HCP) and we acquired ex vivo T1w/T2w. HCP data was already preprocessed according myelin staining from four post-mortem callosal samples. to the HCP preprocessing pipeline (described in Van Essen Additional quantitative susceptibility mapping (QSM) (Li et al. 2013), and made available also with the correspond- et  al. 2015) and MWF data (MacKay et al. 1994; Zhang ing transformation fields to MNI space. HCP-Lifespan data et al. 2015) were included to complement the MRI analysis. were processed by the authors following the same process- We focused on the corpus callosum because it repre- ing pipeline as available at https:// github. com/ Washi ngton- sents, among all the brain pathways, not only the largest Unive rsity/ Pipel ines. 1 3 Brain Structure and Function Briefly, this preprocessing pipeline involves three distinct approach is currently one of the standard methods to quan- pipelines: (1) PreFreeSurfer pipeline, to generate gradient tify myelin water fraction using MRI, and it is expected to distortion corrected, bias field (B1 inhomogeneities) cor - correlate with myelin density highly. In this study, we used rected T1w and T2w images, followed by brain extraction this atlas to compare MWF values with T1w/T2w maps and and registration to standard space; (2) FreeSurfer pipeline, histology in the callosal regions. to segment the structural volumes according to a specific parcellation, reconstruct white and pial cortical surfaces, Ex vivo myelin quantification: histological staining and apply surface registration to FreeSurfer surface atlas; and, finally, (3) the PostFreeSurfer pipeline, which produces Brain samples the final NIFTI files (including T1w/T2w maps) and GIFTI surface files registered to a surface atlas (Conte69), before We obtained four post-mortem brains at autopsy (with the downsampling and conversion to native space. After these prior informed consent of the person's relatives) from sub- steps, average surface cortical myelin maps were created jects without neurological or psychiatric diseases. To test using HCP Connectome Workbench (as described in Glasser whether we could replicate ex vivo the identical T1w/T2w and Van Essen 2011; Glasser et al. 2013). Similarly, average distribution pattern in the corpus callosum across different T1w/T2w MNI volumes for both HCP-Lifespan and HCP ages, three brains (male, 68 years; female, 73 years; male, were obtained using the computed MNI transformations and 82 years) from the Max Planck Institute for Human Cogni- FSL software package (fsl.fmrib.ox.ac.uk/fsl/fslwiki, Jen- tive and Brain Sciences were collected in Leipzig, Germany. kinson et al. 2012). One additional younger female brain (12 years) was obtained from King’s College Hospital, Department of Clinical Neu- Quantitative susceptibility mapping for iron content ropathology, London, United Kingdom. quantification All samples were formalin-fixed at 4% for more than 4 weeks before manually dissecting the corpus callosum. Data from 12 healthy subjects (70 ± 5 years, 6 female) were The three older brain samples were cryoprotected in a 30% acquired on a 3 T MR Biograph Siemens scanner using a sucrose solution and sectioned with a freezing microtome 12-channel coil. The protocol consisted in a spoiled gradi- (SM2000 R, Leica; Hyrax KS 34, Zeiss) in a sagittal plane ent-echo sequence with flow compensation and the follow - at 25 μm. A similar procedure was followed for the young ing parameters: TE = 19.6 ms, TR = 30 ms, flip angle = 12°, female brain where the sample was first formalin-fixed par - voxel size = 0.5 × 0.5 × 1  mm , matrix size = 478 × 378, 160 affin-embedded (FFPE) and then sectioned with a conven- slices. The Quantitative Susceptibility Maps (QSM) were tional microtome at 7 μm. obtained from phase images using the method described by Li et al. 2015. This approach includes the following steps: Myelin staining: luxol fast blue histology a sparse linear equation and least-squares algorithm-based method to derive an initial estimation of magnetic suscep- The sections were mounted on coated slides and stained for tibility; a fast-quantitative susceptibility mapping method myelin sheaths with an optimised Luxol Fast Blue (LFB) to estimate the susceptibility boundaries and an iterative protocol (Mulisch and Welsch 2015). The deparaffinised approach to estimate the susceptibility artifact from ill- sections were rinsed in distilled water for 5 min, and in 70, conditioned k-space regions only. The procedure generates 85, and 96% alcohol for 3 min each. They were incubated an unbiased estimate of tissue susceptibility with negligible overnight at a temperature of 60 °C in a 0,1% LFB solu- streaking artifacts. QSM maps of each subject were spatially tion (100 ml 96% alcohol, 0,1 g LFB (SERVA) and 0,5 ml normalised to MNI template by applying T1-driven spatial 10% acetic acid), followed by 96% alcohol for 3 min, and non-rigid transformation, estimated with SPM12; QSM distilled water for 5 min. They were subsequently rinsed in maps were voxel-wise averaged across all subjects to obtain 0.05% lithium carbonate for 5 seconds, differentiated two a QSM average map. times in 70% alcohol (30 s each time, with short movements to rinse out the excess colour), and rinsed twice in distilled Multi‑echo T2 myelin water fraction atlas water for 5 minutes. The sections were then dehydrated in graded alcohols, rinsed twice in Toluol and coverslipped An atlas of the average MWF distribution in the adult with Entellan (Merck). human brain has been made available at https://sour cef orge. net/ proje cts/ myelin- water- atlas/ (Liu et al. 2019). In brief, LFB optical density analysis this atlas was made using a multi-echo T2 approach, as described by MacKay et al. 1994, Zhang et al. 2015, using We digitized the sections of the three post-mortem older a 3D GRASE pulse sequence (Prasloski et al. 2012). This callosal samples with a digital camera (D90, Nikon), and 1 3 Brain Structure and Function the younger one with a photomicroscope with a motor- Statistical analysis ized stage (Axio Imager M1, Zeiss; 4 × magnification). We estimated myelin density using an LFB Optical Den- To quantitatively compare MRI maps with histological data, sity analysis approach (Beckmann et al. 2018) with the the same 92 ROIs sampling procedure was applied to the Nikon NIS-Elements Advanced Research 4.12 software. average T1w/T2w and MWF maps. Each map was sampled All acquisitions were made with the same light calibra- on two sagittal slices at + 4 and − 4 mm from the midsagit- tion. Three-channel RGB callosal images were converted tal section, and values were averaged. Histological LFB-OD to a single 256-grey scale. The 0–256 intensity range was data were averaged across the three adult callosal samples. rescaled from 1 to 7 and converted into a colour-coded Pearson correlations were performed to compare 92 ROIs map, where purple indicates the lowest level of myelin LFB-OD vs T1w/T2w, LFB-OD vs MWF and finally T1w/ density and dark red the highest level of myelin density T2w vs MWF. (Deshmukh et al. 2013). Results Definition of the callosal regions of interest (ROIs) Figure 1A illustrates the cortical distribution of the T1w/ A sampling over 92 circular regions of interest (ROIs) was T2w ratio in the 57 subjects from the HCP dataset. This performed following a template of the midsagittal section result coherently replicates previous findings from Glasser of the corpus callosum. As shown in the results section, and Van Essen (2011): high T1w/T2w ratio values can be the template is organised according to the anatomical and found in regions with high cortical myelination, such as geometrical subdivision proposed by Witelson 1989. It the primary motor-sensory and visual cortex. Figure  1B subdivides the corpus callosum into six macro-regions represents the volumetric distribution of T1w/T2w in the along the antero-posterior axis: rostrum/genu, rostral white matter for both the HCP and HCP-Life Span datasets. body, anterior midbody, posterior midbody, isthmus, and The highest T1w/T2w intensities is not in the white matter, splenium. The template is specifically designed based on but, instead, in grey matter regions such as the red nucleus, Witelson’s geometrical subdivision to take into account substantia nigra, globus pallidus and partially the dentate individual variability in callosal size among the subjects nucleus. This result is consistent in both the older and and allow comparisons among them (Witelson 1989). younger cohort. In the white matter of the corpus callosum, Fig. 1 T1w/T2w myelin maps at cortical and callosal level. A T1w/ (age 45–75, bottom). The highest intensity signals are located in the T2w myelin maps at the cortical level of 57 healthy participants red nuclei (RN) and globus pallidus (GP), while moderate and low (aged 21–35) from the Human Connectome Project. T1w/T2w corti- intensities are in the dentate nucleus (DN) and claustrum (CL). Right: cal maps show a gradient of T1w/T2w values from the primary motor averaged T1w/T2w map of the corpus callosum of the same subjects. (M1) and sensory areas (V1, S1, A1), which have higher values, to The colour-code map shows a heterogeneous distribution of T1w/ associative areas. This pattern is consistent with previous reports T2w values in the different callosal subregions, with the highest T1w/ (Nieuwenhuys 2013; Van Essen and Glasser 2014; Nieuwenhuys T2w values in the rostrum/genu and slightly lower in the isthmus/ et  al. 2015). Colour scale: purple, low T1w/T2w values; red, high splenium. Colour scale: purple, low T1w/T2w values; red, high T1w/ T1w/T2w values. B Left: averaged T1w/T2w maps of the whole brain T2w values of healthy participants from HCP (age 21–35, top) and HCP-Lifespan 1 3 Brain Structure and Function T1w/T2w maps reveal a heterogeneous distribution of values across samples of 0.62. An identical pattern, characterised along the callosal subregions, with higher T1w/T2w values by higher myelin content in the medium/posterior part of the in the rostrum/genu and isthmus/splenium compared to the corpus callosum, was also detected in the additional younger posterior midbody and the inferior splenium (Fig. 1B, right). brain sample (female, age: 12 years), which was dissected This pattern was consistently observed in both the HCP and and stained in a different laboratory (Fig.  3). HCP-Life Span datasets, thus suggesting that this intensity To further examine whether T1w/T2w maps are mod- heterogeneity is stable and not driven by age effects. ulated by iron content, we investigated the QSM map To test whether we could replicate ex vivo the identi- derived from an average of 12 healthy subjects (Fig.  4, cal T1w/T2w distribution pattern in the corpus callosum, top). As expected, regions with high QSM intensities we analysed three human post-mortem samples (ages: 68, are the red nucleus, substantia nigra and globus pallidus. 73, 82 years) stained with LFB. LFB staining can be seen These findings align well with what we found in the T1w/ in the histological slices (Fig. 2, centre) and with a 92 cir- T2w maps. However, regions such as the dentate nucleus cular ROIs sampling approach defined along the midsagit- tal sections of the entire corpus callosum (Fig.  2, right). The histological myelin distribution in the corpus callosum derived from LFB-optical density analysis is different from the T1w/T2w pattern. In fact, in all three LFB callosal sam- ples, myelin density in the anterior part (mainly rostrum/ genu) is low and higher in the posterior part (mainly isth- mus/splenium). This staining pattern is reproducible across a series of five consecutive sections for each case (see Sup- plementary Figs. 1–3). Indeed, the observed variance across the callosal regions was lower than the variance within the Fig. 3 Myelin mapping of the corpus callosum: LFB-stained midsag- ittal section of the corpus callosum of a 12-year-old female. Insets: different samples. For each of the 92 callosal data points, we LFB myelin staining at higher magnification shows lower intensity observed a normalised LFB intensity range extending from in the anterior corpus callosum (rostrum/genu, left inset) and higher values of 1 to 7, with an average standard deviation across intensity in the posterior part (isthmus, right inset) all regions of 1.47 but only an average standard deviation Fig. 2 Histology and ROI sampling. Left, top: one of the four post- subjects 1, 2 and 3. Right: ROI-based, colour-coded maps of the mye- mortem brain samples with the corpus callosum in situ. Left, bottom: lin distribution of subjects 1, 2 and 3 along the midsagittal section of template of the corpus callosum showing 92 circular regions of inter- the corpus callosum assessed ex vivo via histological myelin staining. est (ROIs) based on the anatomical subdivision proposed by Witelson The histological myelin distribution pattern of the corpus callosum is (1989). Values were calculated inside the circular ROIs in the centre different from the T1w/T2w pattern. All three LFB callosal samples of each of the 92 squares of the callosal template and then extended have low myelin density in the anterior part (mainly rostrum/genu) to the square surrounding the circle. Centre: Luxol Fast Blue (LFB)- and higher myelin density in the posterior part (mainly isthmus/sple- stained midsagittal sections  (25  µm thick) of the corpora callosa of nium). Colour scale: purple, low values; red, high values 1 3 Brain Structure and Function Fig. 4 QSM and MWF cortical and myelin mapping. Left, top: aver- sity in the dentate nucleus (DN) and claustrum (CL) is not matched in age quantitative susceptibility maps (QSM) illustrating iron distri- the T1w/T2w maps represented in Fig. 1. Left, bottom: Myelin Water bution in the whole brain in 12 healthy subjects (70 ± 5-year-old, 6 Fraction (MWF) maps from (Liu et  al. 2019) averaged MWF atlas. females). Higher intensities are localised in the red nuclei (RN)  and Right: averaged QSM and MWF maps of the corpus callosum show globus pallidus (GP) as in the T1w/T2w maps. The high QSM inten- different intensity patterns from anterior to posterior regions and claustrum have a high QSM intensity, but only moder- To quantitatively evaluate the correlation between T1w/ ate or low T1w/T2w values. At callosal level, QSM val- T2w, MWF and histology, we sampled the two MRI maps ues are high in the inferior part of the genu, rostrum and using the same 92-ROI sampling scheme used for histol- inferior splenium, where T1w/T2w values are also high. ogy. Figure 5 shows the correlation data: T1w/T2w does not However, the upper parts of the genu and splenium have positively correlate with LFB-Optical Density, but, instead, a lower mean magnetic susceptibility, whereas the T1w/ reveals a weak to moderate yet significant negative correla- T2w intensity is higher. A discrepancy is also evident for tion (r = 0.21, p < 0.001, Fig. 5A). On the contrary, MWF the anterior midbody (QSM values very low, T1w/T2w is strongly and positively correlated with LFB (r = 0.52, medium-to-high), whereas the isthmus reveals relatively p < 0.001, Fig. 5B), suggesting a good agreement between low intensities in both QSM and T1w/T2w maps. the two modalities. Finally, T1w/T2w and MWF maps are Compared to the MWF atlas, T1w/T2w maps suggest similarly weakly negatively correlated (r = 0.13, p < 0.001, a similar, increased intensity difference between white Fig. 5C). matter regions and cortical grey matter (Fig. 4, bottom). However, another pattern emerges when looking at differ - ences within white matter regions. The internal capsule Discussion and the region corresponding to the cortico-spinal tract display a high content of MWF, whereas T1w/T2w has an In this study, we applied multiple in vivo MRI and histo- opposite, darker contrast compared to more lateral white logical methods to clarify whether T1w/T2w is a method matter regions. By focusing on the body of the corpus cal- to map myelin in the white matter. Our results showed a losum, the MWF atlas is characterised by relatively high discrepancy between T1w/T2w maps and the alternative myelination in the anterior genu, but lower intensities in imaging modalities we applied. While histological studies the anterior body. Higher MWF intensities are evident have confirmed that cortical regions with high T1w/T2w from the posterior midbody and splenium, thus reproduc- signal co-localise with high myelin levels (Glasser et al. ing a pattern similar to the histological staining intensity. 2014), this contrast does not correlate with alternative 1 3 Brain Structure and Function Fig. 5 To quantitatively evaluate the correlation between T1w/T2w, significant negative correlation (r = 0.21 p < 0.001). B MWF pre- MWF and histology, MRI maps were sampled using the same 92-ROI sents a strong and significant positive correlation with LFB (r = 0.52, sampling scheme used for histology and correlated with the cor- p < 0.001), suggesting a good agreement between the two modalities. responding average LFB-OD values. Data is displayed on arbitrary C T1w/T2w and MWF maps do not correlate positively, as they are units (from 1 to 7) for all methods. A T1w/T2w does not positively characterised by a weak negative correlation (r = 0.13, p < 0.001) correlate with the LFB-Optical Density analysis but, instead, has a methods to quantify myelin in white matter regions. Myelin and iron distribution Overall, the T1w/T2w ratio reveals increased values in white matter regions compared to cortical grey matter, The discrepancy we found at the callosal level suggests that but, within white matter, its intensity variability does other biological factors should be considered when these not follow the same pattern as illustrated by the other maps are applied to the white matter. We adopted LFB as methods. Quite striking is the relatively low T1w/T2w myelin staining, which is not only one of the most used intensity in the regions corresponding to the corticospinal staining methods globally and the reference staining tool tract and internal capsule compared to the other white used in routine diagnostics (Kluver and Barrera 1953; Laz- matter regions with higher intensity. This pattern is not zarini 2003), but it has also been widely used in previous found in the MWF atlas, where the internal capsule and papers to compare myelin mapping MRI methods quantita- the corticospinal have a higher myelin content than white tively and to quantify changes in diseases or animal models matter regions. When mapped against the callosal histol- (Laule et al. 2006; Khodanovich et al. 2017; Beckmann et al. ogy, T1w/T2w seems again to show a different pattern 2018; Wood et al. 2016). compared to LFB-OD analysis. In T1w/T2w, the genu Luxol Fast Blue stain ‘highlights the blue myelinated is the region with the highest ‘myelination’, followed by axons of neurons in the white matter of the nervous system the superior portion of the splenium. On the other hand, and the small dense round nuclei of oligodendrocytes that while indicating an increase in myelination in the most produce this myelin sheath’ (Lindberg and Lamps 2018). anterior part of the genu, LFB permits observing much Myelin and iron distributions co-localise significantly in stronger myelination in the posterior regions of the corpus many regions (Ogg and Steen 1998; Fukunaga et al. 2010), callosum, which is consistent across all the histological especially in the visual cortex and in the motor/somatosen- samples and the subjects, irrespective of age and gen- sory cortex (Stüber et al. 2014), and prior MRI investiga- der. This is further confirmed by the correlation analysis tions of brain iron have been published (Gelman et al. 1999; based on the 92-ROI sampling, where T1w/T2w does not Haacke et al. 2005). Therefore, a plausible alternative expla- positively correlate with the LFB nor the MWF measure- nation is that T1w/T2w maps are modulated by, or might ments. On the contrary, MWF and LFB have a strong reflect, differences in iron density. Our results support this positive correlation (r = 0.72, r = 0.52, p < 0.001), thus hypothesis, as very high T1w/T2w intensities were recorded supporting the idea that both modalities indeed capture in regions that are known for their high iron content, includ- myelin density. These observations suggest that the T1w/ ing the globus pallidus, substantia nigra and red nucleus T2w contrast may not represent a suitable method for (Rouault 2013; Piñero and Connor 2000, but see also Koep- mapping the myelin in the corpus callosum and, arguably, pen 1995). These high-intensity regions are evident in in the white matter in general. both the HCP-Lifespan dataset with older subjects and the 1 3 Brain Structure and Function younger group from HCP, a piece of evidence that seems to is needed to clarify the biological meaning and elucidate the rule out any age effect. T1w/T2w signal's complexity. Our QSM results confirm this overlap with iron at the Other MRI methods have been proposed to quantify mye- level of these subcortical regions. However, this overlap is lin mapping, such as multi-echo-T2 (MacKay et al. 1994; not present for other subcortical regions, such as the den- Zhang et al. 2015), mcDespot (Deoni et al. 2008; Deoni tate nucleus and claustrum. Similarly, relatively high signal and Kolind 2015), quantitative T1 maps (Geyer et al. 2011; intensity is evident for both T1w/T2w and QSM maps in the Geyer and Turner 2013; Bock et al. 2013), and macromo- inferior genu and the inferior splenium, but not in the supe- lecular tissue volume (Mezer et al. 2013). All these methods rior genu, superior splenium and anterior midbody. In light attempt to provide a quantitative measure of myelin, and of this, it seems appropriate to conclude that the iron content their histological validation is currently ongoing (Lazari may modulate the T1w/T2w ratio only in some subcortical and Lipp 2021; Laule et al. 2006; Laule 2008; Dula et al. nuclei, and possibly some subregions of the corpus callo- 2010; Fatemi et al. 2011; Khodanovich et al. 2017). To date, sum, but it does not fully explain the origin of the T1w/T2w whether myelin mapping biomarkers can give absolute and contrast. The lack of published histologically defined iron quantitative indications of myelin density is still an open maps of the corpus callosum does not allow us to directly question. Nevertheless, T1w/T2w maps cannot be fully compare myelin vs. iron distribution to test further hypoth- considered quantitative because they heavily depend on the eses beyond our QSM results. actual scanning parameters (i.e., multiple T1w and T2w con- MWF maps correlate well with histology and seem trasts can be obtained on the same subject by changing the good myelin maps. However, various structures definitively MR acquisition parameters), and this effectively changes the known not to contain myelin also show up with high MWF, final contrast and their reproducibility. Therefore, T1w/T2w including the dura (especially the falx in the publicly avail- maps should be assumed to provide qualitative information, able atlas), the cerebral arteries and veins, and the extraocu- and particular care should be taken when comparing results lar muscles. Moreover, MWF images seem strongly affected across studies or scanner manufacturers. by iron (e.g., in the globus pallidus), similar to T1w/T2w, NODDI, T2w and quantitative T2. Overall, it can be intrigu- Limitations of the study ing to speculate on the potential differences in the iron pres- ence (ferritin, within microglia, etc.) and if and how that This study has some limitations. Our group of older par- may impact QSM and T1/T2 measures differently, although ticipants encompassed ten subjects. However, by comparing this is clearly outside the scope of the paper and will be T1w/T2w maps in a younger cohort of 57 HCP participants, clarified by future works. we obtained identical results, and we concluded that T1w/ T2w findings are not age-dependent. Another limitation is Myelin and axonal diameter that our report is based on an indirect comparison between in vivo and ex vivo maps acquired from different subjects. An alternative hypothesis is that the T1w/T2w signal might Ideally, MRI and histology should have been acquired from also be modulated by the complex distribution of different the same subjects. This is certainly feasible with animal populations of fibres with varying axonal diameters. His- models or by taking advantage of human post-mortem imag- tological maps of the axonal diameter distribution along ing, albeit for the latter the correspondence between in vivo the corpus callosum have been published by Aboitiz et al. T1w/T2w and post-mortem T1w/T2w still needs to be veri- (1992), and their pattern of distribution strikingly overlaps fied, and this would have introduced a further source of with our T1w/T2w maps. In particular, the higher T1w/ variability. Similarly, the QSM map and the MWF atlas are T2w signal in the anterior part of the corpus callosum cor- derived from different groups of individuals. However, as we responds to a region containing small-diameter axons. In detected the same T1w/T2w pattern for different cohorts and contrast, the central segment of the corpus callosum with the same pattern in each histology sample, we expect that a lower T1w/T2w signal corresponds to a region with pre- what is measured in this study is overall robust and beyond dominantly large-diameter axons. Furthermore, the mixed interindividual variability. Nevertheless, an approach that low–high-low intensity pattern in the posterior corpus cal- could be applied in future studies would be to collect QSM, losum parallels the large–small–large pattern of the axonal multi-echo T2-MWF and T1w/T2w on the same group of diameter distribution (Aboitiz et al. 1992). subjects. Overall, at this point, we cannot exclude that the T1w/ We adopted LFB as myelin staining for the reasons T2w signal is a rather aspecific contrast modulated by mul- explained above, but we are aware that other histological tiple biological factors. In fact, in addition to myelin, iron methods are available, including staining for myelin basic and fibre density, other glial cells, elements of extracellular protein (MBP) and myelin proteolipid protein (PLP). Moreo- space and vasculature may play a role. Additional research ver, a quantification of the callosal myelin density has been 1 3 Brain Structure and Function performed in the macaque brain using high-resolution Elec- resolution, which, in turn, might lead to significant partial tron Microscopy (EM) data (Stikov et al. 2015a and b). How- voluming and rather long acquisition time. MWF maps ever, while EM can provide accurate and precise quantifica- correlate well with histology. However, considering the tion of myelin, only small sample regions can be selected. non-myelinated structures displaying high MWF values, If samples and tissues are not homogenous, this may lead it is plausible that these are not entirely specific for myelin to notable sampling bias. Considering that our human sam- either. The quest for the most suitable MRI measure for ples were fixed by immersion, compared to perfusion fixa- quantifying myelin is still open (Mancini et al. 2020), and tion used in animal models, this may have introduced more future studies can potentially extend this finding by using local inhomogeneities, making the final interpretation of the also MT-based methods (Hagiwara et al. 2018; Henkelman results more difficult. et al. 2001). On a more general note, methodological and In this paper, we used the anatomical and geometrical performance variability, inconsistent reporting, publica- subdivision of the corpus callosum proposed by Witelson tion bias and limited validation data available are report- (1989). We are aware that more refined parcellation sub- edly some of the factors limiting the comparison between divisions have been proposed, such as those by Hofer and MRI-based myelin markers (Lazari and Lipp 2021; van der Frahm (2006) and Caminiti et al. (2013), to study the precise Weijden et al. 2021). topography of callosal projections. However, the choice of the Witelson subdivision was driven by the need to (i) com- pare our results with those obtained in the reference work by Aboitiz et al. (1992), (ii) define a reproducible geometrical Conclusions constraint for the 92 regions of interest used to quantify mye- lin density across the whole corpus callosum, (iii) to provide Developing a reliable in vivo method for myelin mapping comparable anatomical information with previous studies is crucial to understanding myelination in healthy and (Aboitiz et al. 1992; Zaidel and Iacoboni 2003; Catani and pathological conditions. This paper represents a prelimi- Thiebaut de Schotten 2012). nary attempt to clarify the biological underpinnings of the Finally, the handling and processing of post-mortem T1w/T2w signal. The results highlight important limita- brains involved many steps that could potentially introduce tions in interpreting T1w/T2w as a valid myelin contrast biases. We cannot exclude that some of the inter-individual in white matter. The discrepancy between T1w/T2w MRI variabilities might be related to differences in slice location maps, QSM, MWF and histological myelin maps sug- or inhomogeneity along the slice due to cutting artefacts or gests caution in using T1w/T2w as a white matter map- vascularization. However, these are unlikely to impact our ping method. Future studies are needed to clarify the exact findings as the same pattern of myelin density was preserved biological meaning and the origin of the T1w/T2w signal and commonly present in all the consecutive stained callosal in the white matter. slices, beyond inter-individual differences and across differ - Supplementary Information The online version contains supplemen- ent laboratories (Supplementary Figs. 1–3). tary material available at https://doi. or g/10. 1007/ s00429- 022- 02600-z . Although the paper focuses on ‘putting to the test’ the T1w/T2w method as a myelin mapping tool, it is worth stat- Funding Stefano Sandrone was funded by the National Institute for Health Research (NIHR) Biomedical Research Centre at South London ing that all the imaging methods have strengths and weak- and Maudsley NHS Foundation Trust and King’s College London. nesses.  For example, T1w/T2w maps’ strengths include Marco Catani is the recipient of a Wellcome Trust Investigator Award ease of acquisition on 3 T scanners, high spatial resolution, (103759/Z/14/Z). The London Neurodegenerative Diseases Brain Bank receives funding from the MRC and as part of the Brains for Demen- high contrast-to-noise ratio (CNR), short acquisition times tia Research programme, jointly funded by Alzheimer’s Research UK (Glasser and Van Essen 2011). Furthermore, T1w/T2w and Alzheimer’s Society. Michel Thiebaut de Schotten is funded by images are useful for other kinds of processing, including the European Research Council (ERC) under the European Union’s subcortical and cortical segmentation or surface generation Horizon 2020 research and innovation programme (Grant agreement No. 818521). Flavio Dell’Acqua is funded by the Sackler Institute for (Glasser et al. 2014). At the same time, these are not quan- Translational Neurodevelopment. Michel Thiebaut de Schotten would titative, and there may be a second order anti-correlation like to thank the University of Bordeaux−s IdEx “Investments for the with other white matter measures related to the predominant Future” program RRI “IMPACT,” which received financial support axonal diameter in the white matter. from the French government. We thank the staff of the Clinical Neuro- pathology Department at King’s College Hospital for diagnostic and Moreover, these are typically harder to acquire at the technical assistance, and we also thank the donors and their families. same acquisition resolution, field strength, acquisition time, and CNR as T1w/T2w. MWF images are closer Data availability This manuscript has no associated data. to the histological profile, as shown here. Yet the pub- licly available datasets seem to have a very low spatial 1 3 Brain Structure and Function A regenerative approach to the treatment of multiple sclerosis. Declarations Nature 502(7471):327–332 Dula AN, Gochberg DF, Valentine HL, Valentine WM, Does MD Conflict of interest Stefano Sandrone has no relevant financial or non- (2010) Multiexponential T2, magnetization transfer, and quan- financial interests to disclose. titative histology in white matter tracts of rat spinal cord. Magn Reson Med 63(4):902–909 Open Access This article is licensed under a Creative Commons Attri- Fatemi A, Wilson MA, Phillips AW, McMahon MT, Zhang J, Smith bution 4.0 International License, which permits use, sharing, adapta- SA, Arauz EJ, Falahati S, Gummadavelli A, Bodagala H, Mori S, tion, distribution and reproduction in any medium or format, as long Johnston MV (2011) In vivo magnetization transfer MRI shows as you give appropriate credit to the original author(s) and the source, dysmyelination in an ischemic mouse model of periventricular provide a link to the Creative Commons licence, and indicate if changes leukomalacia. J Cereb Blood Flow Metab 31(10):2009–2018 were made. The images or other third party material in this article are Fukunaga M, Li TQ, van Gelderen P, de Zwart JA, Shmueli K, included in the article's Creative Commons licence, unless indicated Yao B, Lee J, Maric D, Aronova MA, Zhang G, Leapman RD, otherwise in a credit line to the material. If material is not included in Schenck JF, Merkle H, Duyn JH (2010) Layer-specific variation the article's Creative Commons licence and your intended use is not of iron content in cerebral cortex as a source of MRI contrast. permitted by statutory regulation or exceeds the permitted use, you will Proc Natl Acad Sci U S A 107(8):3834–3839 need to obtain permission directly from the copyright holder. To view a Gelman N, Gorell JM, Barker PB, Savage RM, Spickler EM, Wind- copy of this licence, visit http://cr eativ ecommons. or g/licen ses/ b y/4.0/ . ham JP, Knight RA (1999) MR imaging of human brain at 3.0 T: preliminary report on transverse relaxation rates and relation to estimated iron content. Radiology 210(3):759–767 References Geyer S, Turner R (2013) Microstructural parcellation of the human cerebral cortex. Springer-Verlag, Berlin Geyer S, Weiss M, Reimann K, Lohmann G, Turner R (2011) Micro- Aboitiz F, Scheibel AB, Fisher RS, Zaidel E (1992) Fiber composi- structural parcellation of the human cerebral cortex-From Brod- tion of the human corpus callosum. Brain Res 598(1–2):143–153 mann’s post-mortem map to in vivo mapping with high-field Alberts B, Alberts B, Johnson A, Lewis J, Raff M, Roberts K, Walter P magnetic resonance imaging. Front Hum Neurosci 5:19 (2014) Molecular biology of the cell. Garland Science, New York Glasser MF, Van Essen DC (2011) Mapping human cortical Arshad M, Stanley JA, Raz N (2017) Test–retest reliability and con- areas in vivo based on myelin content as revealed by T1 and current validity of in vivo myelin content indices: myelin water T2-weighted MRI. J Neurosci 31(32):11597–11616 fraction and calibrated T1w/T2w image ratio. Hum Brain Mapp Glasser MF, Goyal MS, Preuss TM, Raichle ME, Van Essen DC 38(4):1780–1790 (2014) Trends and properties of human cerebral cortex: correla- Beckmann N, Giorgetti E, Neuhaus A et al (2018) Brain region-specific tions with cortical myelin content. Neuroimage 93(2):165–175 enhancement of remyelination and prevention of demyelination Glasser MF, Sotiropoulos SN, Wilson JA, Coalson TS, Fischl B, by the CSF1R kinase inhibitor BLZ945. Acta Neuropathol Com- Andersson JL, Xu J, Jbabdi S, Webster M, Polimeni JR, Van mun 6(1):9 Essen DC, Jenkinson M, WU-Minn HCP (2013) The minimal Bock NA, Hashim E, Janik R, Konyer NB, Weiss M, Stanisz GJ, Turner preprocessing pipelines for the human connectome project. R, Geyer S (2013) Optimizing T1-weighted imaging of cortical Neuroimage 15(80):105–124 myelin content at 3.0 T. Neuroimage 65:1–12 Haacke EM, Cheng NY, House MJ, Liu Q, Neelavalli J, Ogg RJ, Caminiti R, Ghaziri H, Galuske R, Hof PR, Innocenti GM (2009) Khan A, Ayaz M, Kirsch W, Obenaus A (2005) Imaging iron Evolution amplified processing with temporally dispersed slow stores in the brain using magnetic resonance imaging. Magn neuronal connectivity in primates. Proc Natl Acad Sci USA Reson Imaging 23(1):1–25 106(46):19551–19556 Hagiwara A, Hori M, Kamagata K, Warntjes M, Matsuyoshi D, Caminiti R, Carducci F, Piervincenzi C, Battaglia-Mayer A, Confa- Nakazawa M, Ueda R, Andica C, Koshino S, Maekawa T, Irie lone G, Visco-Comandini F, Pantano P, Innocenti GM (2013) R, Takamura T, Kumamaru KK, Abe O, Aoki S (2018) Myelin Diameter, length, speed, and conduction delay of callosal axons measurement: comparison between simultaneous tissue relax- in macaque monkeys and humans: comparing data from histol- ometry, magnetization transfer saturation index, and T1w/T2w ogy and magnetic resonance imaging diffusion tractography. J ratio methods. Sci Rep 8(1):1–12 Neurosci 33(36):14501–14511 Henkelman RM, Stanisz GJ, Graham SJ (2001) Magnetization trans- Catani M, Thiebaut de Schotten M (2012) Atlas of human brain con- fer in MRI: a review. NMR Biomed 14(2):57–64 nections. Oxford University Press, Oxford Hofer S, Frahm J (2006) Topography of the human corpus callosum Chen H, Budin F, Noel J, Prieto JC, Gilmore J, Rasmussen J, Wad- revisited-comprehensive fiber tractography using diffusion ten- hwa PD, Entringer S, Buss C, Styner M (2017) White mat- sor magnetic resonance imaging. Neuroimage 32(3):989–994 ter fiber-based analysis of T1w/T2w ratio map. Med Imag Jenkinson M, Beckmann CF, Behrens TEJ, Woolrich MW, Smith SM 2017(10133):195–201 (2012) FSL. Neuroimage 62(2):782–790 Colmenares AM, Voss MW, Fanning J, Salerno EA, Gothe NP, Kandel ER, Schwartz JH, Jessell TM, Siegelbaum SA, Hudspeth AJ Thomas ML, Edward McAuley E, Kramer AF, Burzynska AZ (2012) Principles of neural science. McGraw-Hill Professional, (2021) White matter plasticity in healthy older adults: the effects New York of aerobic exercise. Neuroimage 239:118305 Khodanovich MY, Sorokina IV, Glazacheva VY et al (2017) Histologi- Deoni SC, Kolind SH (2015) Investigating the stability of mcDESPOT cal validation of fast macromolecular proton fraction mapping as a myelin water fraction values derived using a stochastic region quantitative myelin imaging method in the cuprizone demyelina- contraction approach. Magn Reson Med 73(1):161–169 tion model. Sci Rep 7:46686 Deoni SC, Rutt BK, Arun T, Pierpaoli C, Jones DK (2008) Gleaning Kluver H, Barrera E (1953) A method for the combined staining of multicomponent T1 and T2 information from steady-state imaging cells and fibers in the Nervous system. J Neuropath Exp Neurol data. Magn Reson Med 60(6):1372–1387 12:400–403 Deshmukh VA, Tardif V, Lyssiotis CA, Green CC, Kerman B, Kim Koeppen AH (1995) The history of iron in the brain. J Neurol Sci HJ, Padmanabhan K, Swoboda JG, Ahmad I, Kondo T, Gage FH, 134(Suppl):1–9 Theofilopoulos AN, Lawson BR, Schultz PG, Lairson LL (2013) 1 3 Brain Structure and Function Laule C (2008) Myelin water imaging of multiple sclerosis at 7 T: Rouault TA (2013) Iron metabolism in the CNS: implications for neu- correlations with histopathology. Neuroimage 40(4):1575–1580 rodegenerative diseases. Nat Rev Neurosci 14(8):551–564 Laule C, Leung E, Lis DK et al (2006) Myelin water imaging in multi- Schmahmann JD, Pandya DN (2006) Fiber pathways of the brain. ple sclerosis: quantitative correlations with histopathology. Mult Oxford University Press, New York Scler 12(6):747–753 Shafee R, Buckner RL, Fischl B (2015) Gray matter myelination of Lazari A, Lipp I (2021) Can MRI measure myelin? Systematic review, 1555 human brains using partial volume corrected MRI images. qualitative assessment, and meta-analysis of studies validating Neuroimage 105:473–485 microstructural imaging with myelin histology. Neuroimage Stanisz GJ, Kecojevic A, Bronskill MJ, Henkelman RM (1999) Char- 230:117744 acterizing white matter with magnetization transfer and T2. Mag Lazzarini RA (2003) Myelin biology and disorders. Elsevier Academic Res Med 42(6):1128–1136 Press, San Diego Stikov N, Campbell JS, Stroh T et al (2015a) Quantitative analysis Lee K, Cherel M, Budin F, Gilmore J, Consing KZ, Rasmussen J, of the myelin g-ratio from electron microscopy images of the Wadhwa PD, Entringer S, Glasser MF, Van Essen DC, Buss C, macaque corpus callosum. Data Brief 4:368–373 Styner M (2015) Early postnatal myelin content estimate of white Stikov N, Campbell JS, Stroh T et al (2015b) In vivo histology of the matter via T1w/T2w ratio. Med Imag 9417:484–490 myelin g-ratio with magnetic resonance imaging. Neuroimage Li W, Wang N, Yu F, Han H, Cao W, Romero R, Tantiwongkosi B, 118:397–405 Duong TQ, Liu C (2015) A method for estimating and removing Stüber C, Morawski M, Schäfer A, Labadie C, Wähnert M, Leuze streaking artifacts in quantitative susceptibility mapping. Neuro- C, Streicher M, Barapatre N, Reimann K, Geyer S, Spemann image 108:111–122 D, Turner R (2014) Myelin and iron concentration in the Lindberg MR, Lamps LW (2018) Diagnostic pathology: normal histol- human brain: a quantitative study of MRI contrast. Neuroimage ogy. Elsevier, Amsterdam 93(1):95–106 Liu H, Rubino C, Dvorak AV, Jarrett M, Ljungberg E, Vavasour IM, Uddin MN, Figley TD, Solar KG, Shatil AS, Figley CR (2019) Com- Lee LE, Kolind SH, MacMillan EL, Traboulsee A, Lang DJ, parisons between multi-component myelin water fraction, T1w/ Rauscher A, Li DKB, MacKay AL, Boyd LA, Kramer JLK, Laule T2w ratio, and diffusion tensor imaging measures in healthy C (2019) Myelin water atlas: a template for myelin distribution in human brain structures. Sci Rep 9(1):1–17 the brain. J Neuroimaging 29(6):699–706 Uddin MN, Figley TD, Marrie RA, Figley CR, CCOMS Study Group MacKay A, Whittall K, Adler J, Li D, Paty D, Graeb D (1994) In Vivo (2018) Can T1w/T2w ratio be used as a myelin-specific meas - visualization of myelin water in brain by magnetic resonance. ure in subcortical structures? Comparisons between FSE-based Magn Reson Med 31:673–677 T1w/T2w ratios, GRASE-based T1w/T2w ratios and multi-echo Mancini M, Karakuzu A, Cohen-Adad J, Cercignani M, Nichols TE, GRASE-based myelin water fractions. NMR Biomed 31(3):e3868 Stikov N (2020) An interactive meta-analysis of MRI biomarkers van der Knaap MS, Valk J (2011) Magnetic resonance of myelination of myelin. Elife 9:e61523 and myelin disorders. Springer, Berlin Mezer A, Yeatman JD, Stikov N, Kay KN, Cho NJ, Dougherty RF, van der Weijden CWJ, García DV, Borra RJH, Thurner P, Meilof JF, Perry ML, Parvizi J, le Hua H, Butts-Pauly K, Wandell BA van Laar PJ, Dierckx RAJO, Gutmann IW, de Vries EFJ (2021) (2013) Quantifying the local tissue volume and composition in Myelin quantification with MRI: a systematic review of accuracy individual brains with magnetic resonance imaging. Nat Med and reproducibility. Neuroimage 226:117561 19(12):1667–1672 Van Essen DC, Smith SM, Barch DM, Behrens TE, Yacoub E, Ugurbil Mulisch M, Welsch U (2015) Romeis mikroskopische technik. Springer K, WU-Minn HCP Consortium (2013) The WU-Minn Human Spektrum, Berlin Connectome project: an overview. Neuroimage 15(80):62–79 Nieuwenhuys R (2013) The myeloarchitectonic studies on the human Van Essen DC, Glasser MF (2014) In vivo architectonics: a cortico- cerebral cortex of the Vogt-Vogt school, and their significance for centric perspective. Neuroimage 93(2):157–164 the interpretation of functional neuroimaging data. Brain Struct Witelson SF (1989) Hand and sex differences in the isthmus and genu Funct 218(2):303–352 of the human corpus callosum. Postmortem Morphol Study Brain Nieuwenhuys R, Broere CA, Cerliani L (2015) A new myeloarchitec- 112(3):799–835 tonic map of the human neocortex based on data from the Vogt- Wood TC, Simmons C, Hurley SA et al (2016) Whole-brain ex-vivo Vogt school. Brain Struct Funct 220(5):2551–2573 quantitative MRI of the cuprizone mouse model. PeerJ 4:e2632 Nieuwenhuys R, Voogd J, van Huijzen C (2007) The human central Zaidel E, Iacoboni M (2003) The parallel brain. the cognitive neurosci- nervous system: a synopsis and atlas. Steinkopff, Amsterdam ence of the corpus callosum Cambridge. MIT University Press, Ogg RJ, Steen RG (1998) Age-related changes in brain T1 are corre- Cambridge lated with iron concentration. Magn Reson Med 40(5):749–753 Zhang J, Kolind SH, Laule C, MacKay AL (2015) Comparison of Piñero DJ, Connor JR (2000) Iron in the brain: an important contrib- myelin water fraction from multiecho T2 decay curve and steady- utor in normal and diseased states. Neuroscientist 6(6):435–453 state methods. Magn Reson Med 73(1):223–232 Prasloski T, Rauscher A, MacKay AL, Hodgson M, Vavasour IM, Laule C, Mädler B (2012) Rapid whole cerebrum myelin Publisher's Note Springer Nature remains neutral with regard to water imaging using a 3D GRASE sequence. Neuroimage jurisdictional claims in published maps and institutional affiliations. 63(1):533–539 Quarles RH, Macklin WB, Morell P (2006) Myelin formation, struc- ture and biochemistry. In: Siegel GJ, Albers RW, Brady ST, Price D (eds) Basic Neurochemistry: molecular, cellular and medical aspects. Academic Press Elsevier, New York 1 3

Journal

Brain Structure and Function – Springer Journals

Published: Mar 1, 2023

Keywords: Corpus callosum; Myelin mapping; Neuroanatomy; Neuroimaging; Validation; Quantitative susceptibility mapping; Myelin water fraction

There are no references for this article.