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

Learn More →

Virtual monoenergetic images from photon-counting spectral computed tomography to assess knee osteoarthritis

Virtual monoenergetic images from photon-counting spectral computed tomography to assess knee... Background: Dual-energy computed tomography has shown a great interest for musculoskeletal pathologies. Photon-counting spectral computed tomography (PCSCT) can acquire data in multiple energy bins with the potential to increase contrast, especially for soft tissues. Our objectives were to assess the value of PCSST to characterise cartilage and to extract quantitative measures of subchondral bone integrity. Methods: Seven excised human knees (3 males and 4 females; 4 normal and 3 with osteoarthritis; age 80.6 ± 14 years, mean ± standard deviation) were scanned using a clinical PCSCT prototype scanner. Tomographic image reconstruction was performed after Compton/photoelectric decomposition. Virtual monoenergetic images were generated from 40 keV to 110 keV every 10 keV (cubic voxel size 250 × 250 × 250 μm ). After selecting an optimal virtual monoenergetic image, we analysed the grey level histograms of different tissues and extracted quantitative measurements on bone cysts. Results: The optimal monoenergetic images were obtained for 60 keV and 70 keV. Visual inspection revealed that these images provide sufficient spatial resolution and soft-tissue contrast to characterise surfaces, disruption, calcification of cartilage, bone osteophytes, and bone cysts. Analysis of attenuation versus energy revealed different energy fingerprint according to tissues. The volumes and numbers of bone cyst were quantified. Conclusions: Virtual monoenergetic images may provide direct visualisation of both cartilage and bone details. Thus, unenhanced PCSCT appears to be a new modality for characterising the knee joint with the potential to increase the diagnostic capability of computed tomography for joint diseases and osteoarthritis. Keywords: Bone cysts, Cartilage, Osteophyte, Osteoarthritis (knee), Tomography (X-ray computed) Key points characterise surfaces, disruption, calcification of car- tilage, bone osteophytes, and bone cysts. Photon-counting spectral compute tomography  On the PCSCT virtual monoenergetic images at 60 (PCSCT) is a new tool to explore joints with high keV, volume and density of bone cysts can be spatial resolution. quantified semiautomatically. Virtual monoenergetic images at 60 keV and 70 keV provided sufficient soft-tissue contrast to Background Osteoarthritis (OA) is a chronic inflammatory joint dis- order characterised by cartilage loss, abnormal subchon- dral bone, osteophyte formation, degeneration of * Correspondence: christine.chappard@inserm.fr 1 ligaments, meniscus, and hypertrophy of the capsule [1]. B3OA, CNRS UMR 7052, U 1271 Inserm, University of Paris, Paris, France Full list of author information is available at the end of the article The damage to the cartilage is typically characterised by © The Author(s) under exclusive licence to European Society of Radiology. 2022 Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. Chappard et al. European Radiology Experimental (2022) 6:10 Page 2 of 10 fissures in the superficial layers, which gradually extend possibility to obtain virtual monoenergetic images, with to deeper zones, and finally full cartilage loss [2]. It is es- the potential to increase the contrast especially for soft sential to detect early changes during the reversible tissues [18]. This feature and the reduced detector pixel phase of the disease. Progresses in OA management re- size provided by PCD makes PCSCT a new candidate quire the development of noninvasive diagnostic for medical applications where resolution and soft tissue methods that can be used to quantify changes in the car- contrast are critical. Considering its applications to OA tilage and subchondral bone [3]. to date, PCSCT coupled to an iodine contrast agent has Imaging methods used for diagnosis of OA are usually only been considered in one study, for measuring pro- conventional radiography and magnetic resonance im- teoglycan content in cartilage in knee specimens [19]. aging, the latter being classically used in clinical routine Thus, there are no current methods that have suffi- to visualise joint effusion, cartilage, ligaments, tendons, ciently high resolution and image quality to visualise the meniscus, osteophytes, and bone marrow oedema. internal structures of the bone, the meniscus, and cartil- Computed tomography (CT) is not ideally suited to age details at the same time [20]. the observation of soft tissue like cartilage, however Thus, the aim of this study was to investigate the feasi- when combined with intra-articular application of a con- bility of PCSCT for assessment of joint integrity without trast agent, it can be used to investigate the cartilage the need for a contrast agent. Our first aim is to evaluate surface. Moreover, CT has been proposed for chondro- whether virtual monoenergetic reconstructions produced calcinosis diagnosis [4], and for quantification of both sufficient spatial resolution and soft tissue contrast to volume and number of subchondral bone cysts [5]. visualise both cartilage and bone with enough details Dual-energy CT, combining measurements from two without contrast agent. Our second aim is to evaluate energy spectra has been shown for its use in musculo- the quality of the images produced is sufficient to per- skeletal imaging particularly in the detection of gout [6]. form quantitative analysis. However, a recent publication showed that this tech- nique does not bring more information than conven- Methods tional CT for calcium pyrophosphate deposition [7], and Sample description to the best of our knowledge, it is not used for cartilage Seven knee specimens from 3 males and 4 females, aged analysis. Today, the new generations of photon-counting 80.6 ± 14 years (mean ± standard deviation) were ob- spectral CT (PCSCT) scanners include energy- tained from the Institut d’Anatomie Paris (France). The discriminating photon-counting detectors (PCDs) that collection of these human tissue specimens was con- can simultaneously count photons and resolve their en- ducted according to the relevant protocols established ergy [8, 9] conversely to conventional CT scanners by the Human Ethics Committee from the Institute of energy-integrating detectors. With such systems, it is Medical Research. No additional information was avail- possible to reconstruct different types of images like ma- able regarding cause of death, previous illnesses, or any terial decomposition images or virtual monoenergetic medical treatments of these subjects except for an ab- images. Another advantage of this new detector is the sence of hepatitis and human immunodeficiency virus. improved signal-to-noise ratio (SNR), due to the exclu- The protocol was approved by the French Ministry of sion of electronic noise [10]. With such additional infor- Higher Education and Research (CODECOH number mation, PCSCT is expected to surpass conventional CT DC-2019-3422). After soft tissue removal, the knee spec- and open new possibilities in medical diagnosis. imens were stored at -20 °C. PCSCT is ideal for material decomposition imaging, as for K-edge imaging, which uses the discontinuity at diag- Data acquisition and image reconstruction nostic energies of the linear attenuation coefficient of The knee specimens were imaged using a clinical PCSC high-Z element-based contrast agents, such as gadolin- T prototype system (Philips Healthcare, Amsterdam, ium, gold, and bismuth [11, 12]. Indeed, PCSCT com- Netherland) installed at CERMEP, Lyon. This is a modi- bined with one or several contrast agents has been fied clinical system that is equipped with a conventional proposed for diverse applications. These included track- X-ray tube that can be set to tube voltages from 80 to ing and monitoring of the biodistribution of gold nano- 120 kVp, and tube currents between 10 mA and 500 particles in vivo [13], determination of contrast agent mA; it was set at 120 kVp and 100 mA in the present concentrations in the liver [14], and evaluation of the study. The tube filtration absorbs low-energy X-rays, so risk of breast cancer [15]. PCSCT might also be used to the final spectrum ranges from 30 keV to 120 keV. The quantify calcium content (e.g., in bone, teeth, kidney system is based on PCDs of 2-mm-thick cadmium zinc stones, coronary plaques) and to discriminate between telluride, with a pixel pitch of 270 × 270 μm at the iso- different calcium crystals [16, 17]. In addition to material center, and coupled with application-specific integrated decomposition and K-edge imaging, PCSCT offers the circuits (ChromAIX2, Philips Research Europe, Aachen, Chappard et al. European Radiology Experimental (2022) 6:10 Page 3 of 10 Germany) that operate in single-photon-counting mode and cortical bone and soft tissue with energy for all with energy discrimination [21]. The acquisition time knees. was less than 5 min. The PCDs allow up to five consecu- The second part of the image analysis consisted of the tive energy thresholds between 30 keV and 120 keV, qualitative assessment of OA from the monoenergetic which were set in the present study at 30, 51, 62, 72, and images at the selected energy. For this, the feasibility of 81 keV. The acquisition field of view was 500 mm in- visual assessment of cartilage defects, cartilage calcifica- plane, with a z-coverage of 17.5 mm in the scanner iso- tion, bone cysts, and bone osteophytes was investigated center. Axial scans were performed over 360° with 2,400 by an expert radiologist who had been in practice for > projections per rotation, on a grid of 64 × 1,848 pixels. 20 years (J.B.P.). Fifty stacks of eight slices were acquired to cover the The third part concerned the extraction of quantitative entire knee specimens (height, 10 cm). After data acqui- parameters from the monoenergetic images at the se- sition, the projections in the different energy bins were lected energy. We analysed bone cysts observed in the decomposed on a Compton/photoelectric basis using the monoenergetic images for the two severe OA knee spec- maximum-likelihood method [11]. Then, the decom- imens. To this aim, we performed the segmentation of posed Compton/photoelectric sinograms (Radon trans- the subchondral bone of the femur and patella in the form) were reconstructed using filtered back-projection. femoro-patellar compartments to a total depth of 10 The whole reconstructed images were made of 640 × mm from the subchondral surface. The bone cysts were 640 × 400 voxels, with a voxel size of 250 × 250 × 250 segmented using the commercially available Avizo 9.0 μm with a reconstructed scan field of view about 160 × software (FEI Visualization Sciences Group, Burlington 160 × 100 mm . Seven virtual monoenergetic images MA, Avizo v.9.0) with the semiautomatic region growing from 40 keV to 110 keV were then computed from the tool (Magic Wand). Here, a seed point was first defined linear combination of the reconstructed Compton/ by the operator, and all of the connected voxels with photoelectric images and expressed in Hounsfield units grey levels in a given tolerance range were selected. Any (HU) units. In addition, the conventional HU image ob- object with a volume < 20 voxels (0.31 mm ) was con- tained by combining all of the bins together was sidered as noise, and was removed from the final calcu- computed. lations. The bone cysts in the medial and lateral For comparison, we used a standard HR-pQCT im- compartments were analysed separately. After segmenta- aging protocol (Scanco Wangen-Brüttisellen, tion, the following parameters were measured: number 3 3 Switzerland) with voxel size of 82 μm for all specimens, of cysts/mm , total cyst volume (mm ), and maximum and synchrotron radiation monochromatic CT at 55 cyst volume (mm ). Our quantitative analysis was ex- KeV (European Synchrotron Radiation Facility, Grenoble ploratory, so not supported by statistical analysis. beamline ID 17) with a voxel size at 45 μm for 5 speci- mens. The Kellgren-Lawrence classification [22] was Results performed on the HR-pQCT images. Multienergy imaging Details from one normal specimen (sample 3) are pre- sented here for the reconstructed images from the Image analysis decomposed photoelectric and Compton images, the First, we investigated the multienergy feature of PCSCT. conventional HU image (from merging of all of the en- For this purpose, we analysed the monoenergetic images ergy bins), and the virtual monoenergetic images, from and compared them to the conventional HU images to 40 keV to 110 keV (Fig. 1). We note that the cartilage is select an optimal monoenergetic image for cartilage as- visible in the Compton and conventional HU images, sessment. This required selection of the energy that led and in the virtual monoenergetic images above 50 keV. to the best characterisation of the cartilage in terms of High-energy monoenergetic images above 80 KeV ap- noise and contrast-to-noise ratio (CNR). The noise was peared to provide not only improved contrast, but also computed as the standard deviation (SD) for a manually higher noise. Among all of these monoenergetic images, selected circular region of interest with radius of 4 pixels those for 60 keV and 70 keV appeared less noisy and in a homogeneous part within the cartilage. The CNR with greater image quality than the Compton and con- was computed as the difference for the grey levels along ventional HU images. a line inside the cartilage and a line inside the joint space The noise was computed as the SD inside the cartilage, outside the meniscus divided by the SD previously com- and the CNR as the differences for the grey levels be- puted. The monoenergetic approach allows to retrieve tween the cartilage and the surrounding soft tissue for the information that links the X-ray attenuation to the all of the monoenergetic images and the conventional type of material crossed and the energy. We analysed images (Fig. 2). Among the monoenergetic images, 60 the attenuation variation of cartilage, of both trabecular keV showed the lowest noise levels and the highest Chappard et al. European Radiology Experimental (2022) 6:10 Page 4 of 10 Fig. 1 Details of the reconstructed images for the decomposed materials (photoelectric and Compton effects), conventional image (merging all energy bins), and virtual monoenergetic images (as indicated) for one normal specimen. Orange arrows, upper and lower surfaces of cartilage Fig. 2. Noise standard deviation (SD) (left) and contrast-to-noise ratio (CNR) (right) for the cartilage for all of the monoenergetic images (40−110 keV) and the conventional image (HU). Plots display median and confidence intervals across the seven samples, with red crosses to indicate outliers. Top: Region-of-interest masks used for the analysis. Noise is computed as the SD within the circular region in the cartilage. CNR is computed as the contrast given by black lines 1 and 2 divided by the SD Chappard et al. European Radiology Experimental (2022) 6:10 Page 5 of 10 CNR, where the noise was a little lower than that at 70 specimen (right). Subchondral bone cysts segmentation keV. Adopting the conventional HU image as reference, of the OA specimen in the femur (sample 2) is also dis- the 60 keV monoenergetic image led to a 45% reduction played on Fig. 5. in noise and 75% increase in CNR, which is relevant for PCSCT allowed the visualisation of the cartilage with a good visualisation of cartilage. quality approaching synchrotron radiation CT especially Cartilage, soft tissue surrounding bone, and the differ- for the border of cartilage with joint space (Fig. 6). On ent bone segments showed large variations in attenu- the contrary, cartilage was not visible on conventional ation versus energy (Fig. 3). At 60 keV, the attenuation CT such as HR-pQCT with energy integrating detectors, values ranged from 1,551 to 1,581 HU for cartilage and even with high spatial resolution. from 1,447 to 1,503 HU for soft tissue. Quantitative analysis of selected monoenergetic images Qualitative assessments of monoenergetic images Bone cysts were found in two OA specimens (sample 1 The results of application of the Kellgren-Lawrence clas- and sample 2), and were segmented for both the femur sification on the HR-pQCT images are described in and patella in a region up to 10 mm from the cartilage, Table 1. Cartilage defects qualified by a 75% local cartil- to measure the quantitative parameters. The number age height loss, cartilage calcification, bone cysts are vis- and total volume of the bone cysts, and their maximal ible on the 60 keV monoenergetic images (Fig. 4). Bone volume for each knee are given in Table 2. A three- osteophytes for subjects across the different levels of OA dimensional display of the segmented bone cysts is pro- are pointed out (Fig. 4). Table 1 gives the clinical de- vided in Fig. 5 (red) in sample 2. scriptions of all studied knee specimens relative to the cartilage aspects, osteophytes and subchondral bone Discussion cysts. Virtual monoenergetic images can provide direct visual- Figure 5 shows the three-dimensional displays of the isation of the details of the cartilage and bone on the 60 keV monoenergetic images of two selected samples: same image. To the best of our knowledge, this is the left, OA (sample 2); right, normal (sample 3). The top first study to investigate PCSCT for OA without the images include the femur, the tibia, and the patella. In need for any contrast agent. We have shown that the 60 the bottom images, the patella has been numerically re- keV and 70 keV monoenergetic images are optimal in moved to visualise the cartilage surface. Indeed, cartilage terms of noise and CNR, respectively, and that these surface defects and calcifications can be seen for the OA provide superior image quality compared to the HU specimen (left), with no defect apparent for the normal images. Fig. 3 Right: Energy fingerprints of the different joint tissues computed from the virtual monoenergetic images. Attenuation versus energy for the regions of interest shown in the left panel: patella cartilage (blue), soft tissue different from cartilage (red), cortical bone (yellow), and subchondral bone (green). Plot shows mean and standard deviation across subjects for the selected regions of interest Chappard et al. European Radiology Experimental (2022) 6:10 Page 6 of 10 Table 1 Clinical descriptions of the knee specimens in terms of cartilage, osteophytes, and bone cyts Sample Sex Age KL Lateral Medial Cartilage Femur Patella Femur Patella 1 F 89 4 Thin, defects Thin, defects Thin, defects Thin, defects calcifications+ 2 M 89 3 Thin, irregular Thin, irregular Thin, defects Thin, irregular calcifications+ calcifications+ calcifications+ calcifications+ 3 M 59 1 Normal, smooth Normal, smooth Normal, defects Normal, smooth 4 F 90 0 Normal, smooth Thin, defects Normal, smooth Normal, irregular calcifications+ 5 F 81 0 Normal, smooth Normal, smooth Normal, smooth Normal, smooth calcifications+ calcifications+ Calcifications+ 6 F 94 1 Normal, irregular Normal, irregular Normal, smooth Normal, smooth 7 M 63 2 Thin, defects Thin, defects Normal, defects Thin, defects Subchondral bone cysts 1 Large size, numerous Large size, numerous Large size numerous Large size, numerous 2 Middle size Small size Middle size – few numerous few 3 –– – – 4 Middle size Large size – Middle size numerous numerous few 5 –– – – 6 –– – – 7 Small size Small size –– few few Osteophytes 1 Large size Small size Large size Large size 2 – Small size Small size – 3 –– – – 4 –– – – 5 –– – – 6 –– – – 7 Small size Small size Small size Small size In a previous PCSCT study performed in vivo for the conventional detectors, which consequently decreases abdomen [18], 72 keV monoenergetic images showed the CNR. In contrast, for PCDs, the equally weighted en- improved contrast and SNR compared to conventional ergy photons have positive effects on the contrast [9]. images with similar attenuation patterns. In the present We hypothesise that the higher resolution and higher study, the monoenergetic images in the range from 60 CNR in soft tissue for PCSCT with respect to standard keV to 80 keV had lower noise than the conventional CT will lead to superior diagnosis potential for applica- images, with the 60 keV and 70 keV monoenergetic im- tions for which resolution and contrast are critical. In ages showing similar noise levels, although the lowest addition, another major advantage of virtual monoener- noise overall was for 60 keV. These data confirm the im- getic images is that they provide quantitative attenuation provements in the SNR demonstrated in a previous sim- measurements and reduce beam hardening artifacts [23]. ulated study that used a cadmium zinc telluride PCD in Finally, one of the major advantages of PCD detectors is comparison to conventional detectors [21]. Indeed, the the patient’s radiation dose reduction. Indeed, based on intrinsic qualities of PCDs can explain SNR and CNR in- an American College of Radiology accredited phantom, creases: first, the detector counts the number of pulses it was demonstrated that image quality can be main- greater than the preset threshold, which consequently tained with a reduction in the dose of 40 to 60% [24]. eliminates the electronic noise; secondly, high energy Different joint tissues showed specific energy finger- photons contribute more than low energy photons in prints. In particular, no overlap was seen between the Chappard et al. European Radiology Experimental (2022) 6:10 Page 7 of 10 Fig. 4 Transversal views of the selected 60 keV monoenergetic images for six specimens, the specimens 1 and 2 are severe osteoarthritis and sample 7 moderate osteoarthritis with numbers corresponding to samples. Red circle, cartilage calcifications; yellow circle, cartilage with different thicknesses; blue circle, cartilage defects; purple circle, bone cysts of various size; green circle: osteophytes of various size cartilage and the soft tissue surrounding the cartilage. energy monoenergetic images (i.e.,20−80 keV), it is pos- The analysis of the histograms showed that subjects with sible to differentiate different calcium deposits in periph- cartilage defects are shifted in the lower part of the his- eral joints, such as calcium pyrophosphate and calcium tograms and on the contrary the presence of calcifica- apatite [16, 17] In the present study, PCSCT allowed the tions shifted the histogram in the high part. characterisation of bone cysts, which are related to Indeed, it is possible to discriminate calcifications in mechanical stress in OA knees, and consequently this cartilage and the meniscus. CT has been proposed to will be interesting to study further for better understand- evaluate calcium deposition, which is especially useful in ing of the OA process [25] and it will also help to better the early stages of OA [24]. Using a preclinical scanner phenotype such OA patients [26] and cysts formation with high-resolution (i.e., 0.1 mm voxels) and low- following cartilage repair [27]. Indeed, bone cysts Fig. 5 Three-dimensional displays of the 60 keV monoenergetic for two selected samples. a, left, OA (sample 2); a, right normal (sample 3). White arrows show cartilage defects (top) and calcifications (bottom) on OA. b Illustration of segmented bone cysts (in red) with the semiautomatic region growing tool in OA sample (sample 2) Chappard et al. European Radiology Experimental (2022) 6:10 Page 8 of 10 Fig. 6 Synchrotron radiation images at 55 keV, PCSCT monoenergetic images at 60 keV and HR-pQCT images for 2 samples: a normal one (specimen 4); and an osteoarthritic one (specimen 7). The cartilage border with joint space is visible on the monoenergetic images based on synchrotron radiation, taken as references and the virtual PCSCT monoenergetic images. The yellow arrows correspond to cartilage defects Chappard et al. European Radiology Experimental (2022) 6:10 Page 9 of 10 Table 2 Quantitative analysis of the bone cysts for two of the these data suggest that quantitative measurements of specimens bone cysts in terms of numbers and size might lead Specimen Site Compartment Cysts to new biomarkers for better phenotyping of OA. The prototype is operational for clinical research, with fur- Number Volume (mm ) ther studies needed to validate these data on living Total Maximum patients. Sample 1 Femoral Medial 8 185.8 110.6 Lateral 44 355 106.1 Abbreviations CNR: Contrast-to-noise ratio; HU: Hounsfield unit; OA: Osteoarthritis; Total 52 540.8 110.6 PCD: Photon-counting detectors; PCSCT: Photon-counting spectral Patellar Medial 10 105.1 31 computed tomography; SD: Standard deviation; SNR: Signal-to-noise ratio Lateral 12 371 204.9 Authors’ contributions Total 22 476.1 204.9 All authors read, edited, and approved the final version of the manuscript. CC carried out the study concept, design, data acquisition, analysis and Sample 2 Femoral Medial 3 241.8 109.5 interpretation, and manuscript preparation. JA contributed to data Lateral 4 223.2 125.5 acquisition, data analysis, and manuscript preparation. CO contributed to data acquisition and analysis. SSM contributed to data acquisitions. LB Total 7 465 125.5 contributed to data analysis. JBP carried out the CT examination. PD contributed to study concept, design, interpretation. FP carried out the study Patellar Medial 4 13.2 5.5 concept, design, data acquisition, analysis, and manuscript editing. Lateral 11 37.9 17.7 Funding Total 15 51.1 17.7 This project was supported by the ANR project SALTO (ANR-17-CE19-0011- 01), within LabEx PRIMES (ANR-11-LABX-0063) of University de Lyon. It has received funding from the European Union Horizon 2020 Research and quantification were rarely performed from standard CT Innovation Program under Marie Sklodowska-Curie grant agreement N° 701915. The project also received funding from the European Union Horizon [26] and one time from high-resolution peripheral CT 2020 Research and Innovation Program under grant agreement N° 668142. [28]. This work was partly funded by France Life Imaging (grant ANR-11-INBS- The present study is subject to some limitations. 0006) from the French Investissements d’Avenir. The reconstructed images depend on the material de- Availability of data and materials composition and the postprocessing methods used. The datasets used and/or analysed during the current study are available Material decomposition was carried out on a photo- from the corresponding author on reasonable request. electric and Compton basis. Other specific material Declarations basis, such as those based on bone and soft tissue, or even more general basis, are possible [29]. The inves- Ethics approval and consent to participate tigation of other potentially improved basis might lead The study was approved by the Ethics Committee of Paris University, Paris. The tissue donors or their legal guardians provided informed written to further information with regard to the analysis of consent to give their tissue for investigations, in accord with legal clauses. cartilage and calcifications. The material decompos- itionmethodusedwas basedonconventionalmax- Consent for publication Not applicable imum likelihood, without regularisation in a pixel-by- pixel manner [11]. More advances inverse methods Competing interests including regularisation [30–33] or based on deep The authors of this manuscript declare no relationships with any companies whose products or services may be related to the subject matter of the learning approaches [34–36] might lead to improve- article. ments here. The image post-processing here was spe- cifically designed for knee images; it might be further Author details B3OA, CNRS UMR 7052, U 1271 Inserm, University of Paris, Paris, France. improved [37]. Our quantitative analysis is exploratory University of Lyon, INSA-Lyon, CNRS, Inserm, CREATIS UMR 5220, U1206, at this level, to show the potential of PCSCT for OA. Lyon, France. However, more refined methods in combination with Received: 27 September 2021 Accepted: 30 December 2021 a larger number of samples might lead to new bio- markers that translate into more accurate and meaning- ful OA scoring, and consequently to improved diagnosis. References Further experiments carried out in vivo and with more 1. Loeser RF, Goldring SR, Scanzello CR, Goldring MB (2012) Osteoarthritis: a disease of the joint as an organ. Arthritis Rheum 64:1697–1707. https://doi. subjects are needed to validate these findings. org/10.1002/art.34453 Epub 2012 Mar 5 In conclusion, virtual monoenergetic can provide 2. Thysen S, Luyten FP, Lories RJU (2015) Targets, models and challenges in direct visualisation of both cartilage and bone integ- osteoarthritis research. Dis Model Mech 8:17–30. https://doi.org/10.1242/ dmm.016881 rity, which are essential for the characterisation of 3. Berenbaum F (2010) Targeted therapies in osteoarthritis: a systematic review osteoarticular diseases that affect the whole joint, with of the trials on www.clinicaltrials.gov. Best Pract Res Clin Rheumatol 24:107– this demonstration thus focused on OA. In addition, 119. https://doi.org/10.1016/j.berh.2009.08.007 Chappard et al. European Radiology Experimental (2022) 6:10 Page 10 of 10 4. Magarelli N, Amelia R, Melillo N, Nasuto M, Cantatore F, Guglielmi G (2012) 24. Aslund M, Fredenberg E, Telman M, Danielsson M (2010) Detectors for the Imaging of chondrocalcinosis: calcium pyrophosphate dihydrate (CPPD) future of X-ray imaging. Radiat Prot Dosimetry 139:327–333. https://doi. crystal deposition disease -- imaging of common sites of involvement. Clin org/10.1093/rpd/ncq074 Exp Rheumatol 30:118–125 25. McErlain DD, Milner JS, Ivanov TG, Jencikova-Celerin L, Pollmann SI, 5. Burnett WD, Kontulainen SA, McLennan CE et al (2019) Knee osteoarthritis Holdsworth DW (2011) Subchondral cysts create increased intra-osseous patients with more subchondral cysts have altered tibial subchondral bone stress in early knee OA: a finite element analysis using simulated lesions. mineral density. BMC Musculoskelet Disord 20:14. https://doi.org/10.1186/ Bone 48:639–646. https://doi.org/10.1016/j.bone.2010.11.010 s12891-018-2388-9 26. Burnett WD, Kontulainen SA, McLennan CE, Hunter DJ, Wilson DR, Johnston JD (2014) Regional depth-specific subchondral bone density measures in 6. Metzger SC, Koehm M, Wichmann JL, et al (2016) Dual-energy CT in osteoarthritic and normal patellae: in vivo precision and preliminary patients with suspected gouty arthritis: effects on treatment regimen and comparisons. Osteoporos Int 25:1107–1114. https://doi.org/10.1007/s00198- clinical outcome. Acad Radiol 23:267–272. https://doi.org/10.1016/j.acra.201 013-2568-2 5.10.019 27. Gao L, Cucchiarini M, Madry H (2020) Cyst formation in the subchondral 7. Budzik JF, Marzin C, Legrand J, Norberciak L, Becce F, Pascart T (2021) Can bone following cartilage repair. Clin Transl Med 10:e248. Review. https://doi. dual-energy computed tomography be used to identify early calcium org/10.1002/ctm2.248 crystal deposition in the knees of patients with calcium pyrophosphate 28. Kroker A, Zhu Y, Manske SL, Barber R, Mohtadi N, Boyd SK (2017) deposition. Arthritis Rheumatol 73:687–692. https://doi.org/10.1002/art.41 Quantitative in vivo assessment of bone microarchitecture in the human knee using HR-pQCT. Bone 97:43–48. https://doi.org/10.1016/j.bone.2016.12. 8. Taguchi K, Iwanczyk JS (2013) Vision 20/20: Single photon counting X-ray detectors in medical imaging. Med Phys 40:100901. https://doi.org/10.111 29. Bornefalk H (2012) XCOM intrinsic dimensionality for low-Z elements at 8/1.4820371 diagnostic energies. Med Phys 39:654–657. https://doi.org/10.1118/1.36753 9. Si-Mohamed S, Bar-Ness D, Sigovan M, et al (2018) Multicolour imaging with spectral photon-counting CT: a phantom study. Eur Radiol Exp 2:34. 30. Ducros N, Abascal J, Sixou B, Rit S, Peyrin F (2017) Regularization of https://doi.org/10.1186/s41747-018-0063-4 nonlinear decomposition of spectral X-ray projection images. Med Phys 44: 10. Willemink MJ, Persson M, Pourmorteza A, Pelc NJ, Fleischmann D (2018) e174–e187. https://doi.org/10.1002/mp.12283 Photon-counting CT: technical principles and clinical prospects. Radiology 31. Kim K, Ye JC, Worstell W et al (2015) Sparse-view spectral CT reconstruction 289:293–312. https://doi.org/10.1148/radiol.2018172656 using spectral patch-based low-rank penalty. IEEE Trans Med Imaging 34: 11. Schlomka JP, Roessl E, Dorscheid R, et al (2008) Experimental feasibility of 748–760. https://doi.org/10.1109/TMI.2014.2380993 multi-energy photon-counting K-edge imaging in pre-clinical computed 32. Semerci O, Hao N, Kilmer ME, Miller EL (2014) Tensor-based formulation and tomography. Phys Med Biol 53:4031–4047. https://doi.org/10.1088/0031-91 nuclear norm regularization for multienergy computed tomography. Trans 55/53/15/002 Image Process IEEE 23:1678–1693. https://doi.org/10.1109/TIP.2014.2305840 12. Riederer I, Si-Mohamed S, Ehn S, et al (2019) Differentiation between blood 33. Touch M, Clark DP, Barber W, Badea CT (2016) A neural network-based and iodine in a bovine brain -- Initial experience with spectral photon- method for spectral distortion correction in photon counting X-ray CT. Phys counting computed tomography (SPCCT). PLoS One 14:e0212679. https:// Med Biol 61:6132–6153. https://doi.org/10.1088/0031-9155/61/16/6132 doi.org/10.1371/journal.pone.0212679 34. Arridge S, Maass P, Öktem O, Schönlieb C-B (2019) Solving inverse problems 13. Si-Mohamed S, Cormode DP, Bar-Ness D, et al (2017) Evaluation of spectral using data-driven models. Acta Numerica 28:1–174. https://doi.org/10.1017/ photon counting computed tomography K-edge imaging for determination S0962492919000059 of gold nanoparticle biodistribution in vivo. Nanoscale 9:18246–18257. 35. Bussod S, Abascal JFPJ, Ducros N et al (2019) In: Human knee phantom for https://doi.org/10.1039/c7nr01153a spectral CT: validation of a material decomposition algorithm (ed) 2019 IEEE 14. Si-Mohamed S, Tatard-Leitman V, Laugerette A, et al (2019) Spectral photon- 16th International Symposium on Biomedical Imaging (ISBI 2019), pp 1170– counting computed tomography (SPCCT): in-vivo single-acquisition multi- 1173. https://doi.org/10.1109/ISBI.2019.8759192 phase liver imaging with a dual contrast agent protocol. Sci Rep 9:8458. 36. Ravishankar S, Ye JC, Fessler JA (2020) Image reconstruction: from sparsity to https://doi.org/10.1038/s41598-019-44821-z data-adaptive methods and machine learning. Proc IEEE Inst Electr Electron 15. Ding H, Zhao B, Baturin P, Behroozi F, Molloi S (2014) Breast tissue Eng 108:86–109. https://doi.org/10.1109/JPROC.2019.2936204 decomposition with spectral distortion correction: A postmortem study. 37. Abascal JFPJ, Si-Mohamed S, Douek P, Chappard C, Peyrin F (2019) In: A Med Phys 41:101901. https://doi.org/10.1118/1.4894724 sparse and prior based method for 3D image denoising (ed) 2019 27th 16. Kirkbride TE, Raja AY, Müller K, Bateman CJ, Becce F, Anderson NG (2017) European Signal Processing Conference (EUSIPCO), pp 1–5. https://doi.org/1 Discrimination between calcium hydroxyapatite and calcium oxalate using 0.23919/EUSIPCO.2019.8902564 multienergy spectral photon-counting CT. AJR Am J Roentgenol 209:1088– 1092. https://doi.org/10.2214/AJR.17.18394 17. Stamp LK, Anderson NG, Becce F et al (2019) Clinical utility of multi-energy Publisher’sNote spectral photon-counting computed tomography in crystal arthritis. Arthritis Springer Nature remains neutral with regard to jurisdictional claims in Rheumatol 71:1158–1162. https://doi.org/10.1002/art.40848 published maps and institutional affiliations. 18. Große Hokamp N, Gilkeson R, Jordan MK, et al (2019) Virtual monoenergetic images from spectral detector CT as a surrogate for conventional CT images: Unaltered attenuation characteristics with reduced image noise. Eur J Radiol 117:49–55. https://doi.org/10.1016/j.ejrad.2019.05.019 19. Rajendran K, Löbker C, Schon BS, et al (2017) Quantitative imaging of excised osteoarthritic cartilage using spectral CT. Eur Radiol 27:384–392. https://doi.org/10.1007/s00330-016-4374-7 20. Roemer F, Eckstein F, Hayashi D, Guermazi A (2014) The role of imaging in osteoarthritis. Best Pract Res Clin Rheumatol 28:31–60. https://doi.org/10.101 6/j.berh.2014.02.002 21. Steadman R, Herrmann C, Livne A (2017) ChromAIX2: A large area, high count-rate energy-resolving photon counting ASIC for a Spectral CT Prototype. Nucl Instrum Methods Phys Res Sect Accel Spectrometers Detect Assoc Equip 862:18–24. https://doi.org/10.1016/j. nima.2017.05.010 22. Kellgren JH, Lawrence JS (1957) Radiological assessment of osteoarthritis. Ann Rheumat Dis 16:494–502. https://doi.org/10.1136/ard.16.4.494 23. Kalender WA (2014) Dose in X-ray computed tomography. Phys Med Biol 59:R129–R150. https://doi.org/10.1088/0031-9155/59/3/R129 http://www.deepdyve.com/assets/images/DeepDyve-Logo-lg.png European Radiology Experimental Springer Journals

Virtual monoenergetic images from photon-counting spectral computed tomography to assess knee osteoarthritis

Download PDF
10 pages

Loading next page...
 
/lp/springer-journals/virtual-monoenergetic-images-from-photon-counting-spectral-computed-hFZZjcz0cL

References (45)

Publisher
Springer Journals
Copyright
Copyright © The Author(s) under exclusive licence to European Society of Radiology 2022
eISSN
2509-9280
DOI
10.1186/s41747-021-00261-x
Publisher site
See Article on Publisher Site

Abstract

Background: Dual-energy computed tomography has shown a great interest for musculoskeletal pathologies. Photon-counting spectral computed tomography (PCSCT) can acquire data in multiple energy bins with the potential to increase contrast, especially for soft tissues. Our objectives were to assess the value of PCSST to characterise cartilage and to extract quantitative measures of subchondral bone integrity. Methods: Seven excised human knees (3 males and 4 females; 4 normal and 3 with osteoarthritis; age 80.6 ± 14 years, mean ± standard deviation) were scanned using a clinical PCSCT prototype scanner. Tomographic image reconstruction was performed after Compton/photoelectric decomposition. Virtual monoenergetic images were generated from 40 keV to 110 keV every 10 keV (cubic voxel size 250 × 250 × 250 μm ). After selecting an optimal virtual monoenergetic image, we analysed the grey level histograms of different tissues and extracted quantitative measurements on bone cysts. Results: The optimal monoenergetic images were obtained for 60 keV and 70 keV. Visual inspection revealed that these images provide sufficient spatial resolution and soft-tissue contrast to characterise surfaces, disruption, calcification of cartilage, bone osteophytes, and bone cysts. Analysis of attenuation versus energy revealed different energy fingerprint according to tissues. The volumes and numbers of bone cyst were quantified. Conclusions: Virtual monoenergetic images may provide direct visualisation of both cartilage and bone details. Thus, unenhanced PCSCT appears to be a new modality for characterising the knee joint with the potential to increase the diagnostic capability of computed tomography for joint diseases and osteoarthritis. Keywords: Bone cysts, Cartilage, Osteophyte, Osteoarthritis (knee), Tomography (X-ray computed) Key points characterise surfaces, disruption, calcification of car- tilage, bone osteophytes, and bone cysts. Photon-counting spectral compute tomography  On the PCSCT virtual monoenergetic images at 60 (PCSCT) is a new tool to explore joints with high keV, volume and density of bone cysts can be spatial resolution. quantified semiautomatically. Virtual monoenergetic images at 60 keV and 70 keV provided sufficient soft-tissue contrast to Background Osteoarthritis (OA) is a chronic inflammatory joint dis- order characterised by cartilage loss, abnormal subchon- dral bone, osteophyte formation, degeneration of * Correspondence: christine.chappard@inserm.fr 1 ligaments, meniscus, and hypertrophy of the capsule [1]. B3OA, CNRS UMR 7052, U 1271 Inserm, University of Paris, Paris, France Full list of author information is available at the end of the article The damage to the cartilage is typically characterised by © The Author(s) under exclusive licence to European Society of Radiology. 2022 Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/. Chappard et al. European Radiology Experimental (2022) 6:10 Page 2 of 10 fissures in the superficial layers, which gradually extend possibility to obtain virtual monoenergetic images, with to deeper zones, and finally full cartilage loss [2]. It is es- the potential to increase the contrast especially for soft sential to detect early changes during the reversible tissues [18]. This feature and the reduced detector pixel phase of the disease. Progresses in OA management re- size provided by PCD makes PCSCT a new candidate quire the development of noninvasive diagnostic for medical applications where resolution and soft tissue methods that can be used to quantify changes in the car- contrast are critical. Considering its applications to OA tilage and subchondral bone [3]. to date, PCSCT coupled to an iodine contrast agent has Imaging methods used for diagnosis of OA are usually only been considered in one study, for measuring pro- conventional radiography and magnetic resonance im- teoglycan content in cartilage in knee specimens [19]. aging, the latter being classically used in clinical routine Thus, there are no current methods that have suffi- to visualise joint effusion, cartilage, ligaments, tendons, ciently high resolution and image quality to visualise the meniscus, osteophytes, and bone marrow oedema. internal structures of the bone, the meniscus, and cartil- Computed tomography (CT) is not ideally suited to age details at the same time [20]. the observation of soft tissue like cartilage, however Thus, the aim of this study was to investigate the feasi- when combined with intra-articular application of a con- bility of PCSCT for assessment of joint integrity without trast agent, it can be used to investigate the cartilage the need for a contrast agent. Our first aim is to evaluate surface. Moreover, CT has been proposed for chondro- whether virtual monoenergetic reconstructions produced calcinosis diagnosis [4], and for quantification of both sufficient spatial resolution and soft tissue contrast to volume and number of subchondral bone cysts [5]. visualise both cartilage and bone with enough details Dual-energy CT, combining measurements from two without contrast agent. Our second aim is to evaluate energy spectra has been shown for its use in musculo- the quality of the images produced is sufficient to per- skeletal imaging particularly in the detection of gout [6]. form quantitative analysis. However, a recent publication showed that this tech- nique does not bring more information than conven- Methods tional CT for calcium pyrophosphate deposition [7], and Sample description to the best of our knowledge, it is not used for cartilage Seven knee specimens from 3 males and 4 females, aged analysis. Today, the new generations of photon-counting 80.6 ± 14 years (mean ± standard deviation) were ob- spectral CT (PCSCT) scanners include energy- tained from the Institut d’Anatomie Paris (France). The discriminating photon-counting detectors (PCDs) that collection of these human tissue specimens was con- can simultaneously count photons and resolve their en- ducted according to the relevant protocols established ergy [8, 9] conversely to conventional CT scanners by the Human Ethics Committee from the Institute of energy-integrating detectors. With such systems, it is Medical Research. No additional information was avail- possible to reconstruct different types of images like ma- able regarding cause of death, previous illnesses, or any terial decomposition images or virtual monoenergetic medical treatments of these subjects except for an ab- images. Another advantage of this new detector is the sence of hepatitis and human immunodeficiency virus. improved signal-to-noise ratio (SNR), due to the exclu- The protocol was approved by the French Ministry of sion of electronic noise [10]. With such additional infor- Higher Education and Research (CODECOH number mation, PCSCT is expected to surpass conventional CT DC-2019-3422). After soft tissue removal, the knee spec- and open new possibilities in medical diagnosis. imens were stored at -20 °C. PCSCT is ideal for material decomposition imaging, as for K-edge imaging, which uses the discontinuity at diag- Data acquisition and image reconstruction nostic energies of the linear attenuation coefficient of The knee specimens were imaged using a clinical PCSC high-Z element-based contrast agents, such as gadolin- T prototype system (Philips Healthcare, Amsterdam, ium, gold, and bismuth [11, 12]. Indeed, PCSCT com- Netherland) installed at CERMEP, Lyon. This is a modi- bined with one or several contrast agents has been fied clinical system that is equipped with a conventional proposed for diverse applications. These included track- X-ray tube that can be set to tube voltages from 80 to ing and monitoring of the biodistribution of gold nano- 120 kVp, and tube currents between 10 mA and 500 particles in vivo [13], determination of contrast agent mA; it was set at 120 kVp and 100 mA in the present concentrations in the liver [14], and evaluation of the study. The tube filtration absorbs low-energy X-rays, so risk of breast cancer [15]. PCSCT might also be used to the final spectrum ranges from 30 keV to 120 keV. The quantify calcium content (e.g., in bone, teeth, kidney system is based on PCDs of 2-mm-thick cadmium zinc stones, coronary plaques) and to discriminate between telluride, with a pixel pitch of 270 × 270 μm at the iso- different calcium crystals [16, 17]. In addition to material center, and coupled with application-specific integrated decomposition and K-edge imaging, PCSCT offers the circuits (ChromAIX2, Philips Research Europe, Aachen, Chappard et al. European Radiology Experimental (2022) 6:10 Page 3 of 10 Germany) that operate in single-photon-counting mode and cortical bone and soft tissue with energy for all with energy discrimination [21]. The acquisition time knees. was less than 5 min. The PCDs allow up to five consecu- The second part of the image analysis consisted of the tive energy thresholds between 30 keV and 120 keV, qualitative assessment of OA from the monoenergetic which were set in the present study at 30, 51, 62, 72, and images at the selected energy. For this, the feasibility of 81 keV. The acquisition field of view was 500 mm in- visual assessment of cartilage defects, cartilage calcifica- plane, with a z-coverage of 17.5 mm in the scanner iso- tion, bone cysts, and bone osteophytes was investigated center. Axial scans were performed over 360° with 2,400 by an expert radiologist who had been in practice for > projections per rotation, on a grid of 64 × 1,848 pixels. 20 years (J.B.P.). Fifty stacks of eight slices were acquired to cover the The third part concerned the extraction of quantitative entire knee specimens (height, 10 cm). After data acqui- parameters from the monoenergetic images at the se- sition, the projections in the different energy bins were lected energy. We analysed bone cysts observed in the decomposed on a Compton/photoelectric basis using the monoenergetic images for the two severe OA knee spec- maximum-likelihood method [11]. Then, the decom- imens. To this aim, we performed the segmentation of posed Compton/photoelectric sinograms (Radon trans- the subchondral bone of the femur and patella in the form) were reconstructed using filtered back-projection. femoro-patellar compartments to a total depth of 10 The whole reconstructed images were made of 640 × mm from the subchondral surface. The bone cysts were 640 × 400 voxels, with a voxel size of 250 × 250 × 250 segmented using the commercially available Avizo 9.0 μm with a reconstructed scan field of view about 160 × software (FEI Visualization Sciences Group, Burlington 160 × 100 mm . Seven virtual monoenergetic images MA, Avizo v.9.0) with the semiautomatic region growing from 40 keV to 110 keV were then computed from the tool (Magic Wand). Here, a seed point was first defined linear combination of the reconstructed Compton/ by the operator, and all of the connected voxels with photoelectric images and expressed in Hounsfield units grey levels in a given tolerance range were selected. Any (HU) units. In addition, the conventional HU image ob- object with a volume < 20 voxels (0.31 mm ) was con- tained by combining all of the bins together was sidered as noise, and was removed from the final calcu- computed. lations. The bone cysts in the medial and lateral For comparison, we used a standard HR-pQCT im- compartments were analysed separately. After segmenta- aging protocol (Scanco Wangen-Brüttisellen, tion, the following parameters were measured: number 3 3 Switzerland) with voxel size of 82 μm for all specimens, of cysts/mm , total cyst volume (mm ), and maximum and synchrotron radiation monochromatic CT at 55 cyst volume (mm ). Our quantitative analysis was ex- KeV (European Synchrotron Radiation Facility, Grenoble ploratory, so not supported by statistical analysis. beamline ID 17) with a voxel size at 45 μm for 5 speci- mens. The Kellgren-Lawrence classification [22] was Results performed on the HR-pQCT images. Multienergy imaging Details from one normal specimen (sample 3) are pre- sented here for the reconstructed images from the Image analysis decomposed photoelectric and Compton images, the First, we investigated the multienergy feature of PCSCT. conventional HU image (from merging of all of the en- For this purpose, we analysed the monoenergetic images ergy bins), and the virtual monoenergetic images, from and compared them to the conventional HU images to 40 keV to 110 keV (Fig. 1). We note that the cartilage is select an optimal monoenergetic image for cartilage as- visible in the Compton and conventional HU images, sessment. This required selection of the energy that led and in the virtual monoenergetic images above 50 keV. to the best characterisation of the cartilage in terms of High-energy monoenergetic images above 80 KeV ap- noise and contrast-to-noise ratio (CNR). The noise was peared to provide not only improved contrast, but also computed as the standard deviation (SD) for a manually higher noise. Among all of these monoenergetic images, selected circular region of interest with radius of 4 pixels those for 60 keV and 70 keV appeared less noisy and in a homogeneous part within the cartilage. The CNR with greater image quality than the Compton and con- was computed as the difference for the grey levels along ventional HU images. a line inside the cartilage and a line inside the joint space The noise was computed as the SD inside the cartilage, outside the meniscus divided by the SD previously com- and the CNR as the differences for the grey levels be- puted. The monoenergetic approach allows to retrieve tween the cartilage and the surrounding soft tissue for the information that links the X-ray attenuation to the all of the monoenergetic images and the conventional type of material crossed and the energy. We analysed images (Fig. 2). Among the monoenergetic images, 60 the attenuation variation of cartilage, of both trabecular keV showed the lowest noise levels and the highest Chappard et al. European Radiology Experimental (2022) 6:10 Page 4 of 10 Fig. 1 Details of the reconstructed images for the decomposed materials (photoelectric and Compton effects), conventional image (merging all energy bins), and virtual monoenergetic images (as indicated) for one normal specimen. Orange arrows, upper and lower surfaces of cartilage Fig. 2. Noise standard deviation (SD) (left) and contrast-to-noise ratio (CNR) (right) for the cartilage for all of the monoenergetic images (40−110 keV) and the conventional image (HU). Plots display median and confidence intervals across the seven samples, with red crosses to indicate outliers. Top: Region-of-interest masks used for the analysis. Noise is computed as the SD within the circular region in the cartilage. CNR is computed as the contrast given by black lines 1 and 2 divided by the SD Chappard et al. European Radiology Experimental (2022) 6:10 Page 5 of 10 CNR, where the noise was a little lower than that at 70 specimen (right). Subchondral bone cysts segmentation keV. Adopting the conventional HU image as reference, of the OA specimen in the femur (sample 2) is also dis- the 60 keV monoenergetic image led to a 45% reduction played on Fig. 5. in noise and 75% increase in CNR, which is relevant for PCSCT allowed the visualisation of the cartilage with a good visualisation of cartilage. quality approaching synchrotron radiation CT especially Cartilage, soft tissue surrounding bone, and the differ- for the border of cartilage with joint space (Fig. 6). On ent bone segments showed large variations in attenu- the contrary, cartilage was not visible on conventional ation versus energy (Fig. 3). At 60 keV, the attenuation CT such as HR-pQCT with energy integrating detectors, values ranged from 1,551 to 1,581 HU for cartilage and even with high spatial resolution. from 1,447 to 1,503 HU for soft tissue. Quantitative analysis of selected monoenergetic images Qualitative assessments of monoenergetic images Bone cysts were found in two OA specimens (sample 1 The results of application of the Kellgren-Lawrence clas- and sample 2), and were segmented for both the femur sification on the HR-pQCT images are described in and patella in a region up to 10 mm from the cartilage, Table 1. Cartilage defects qualified by a 75% local cartil- to measure the quantitative parameters. The number age height loss, cartilage calcification, bone cysts are vis- and total volume of the bone cysts, and their maximal ible on the 60 keV monoenergetic images (Fig. 4). Bone volume for each knee are given in Table 2. A three- osteophytes for subjects across the different levels of OA dimensional display of the segmented bone cysts is pro- are pointed out (Fig. 4). Table 1 gives the clinical de- vided in Fig. 5 (red) in sample 2. scriptions of all studied knee specimens relative to the cartilage aspects, osteophytes and subchondral bone Discussion cysts. Virtual monoenergetic images can provide direct visual- Figure 5 shows the three-dimensional displays of the isation of the details of the cartilage and bone on the 60 keV monoenergetic images of two selected samples: same image. To the best of our knowledge, this is the left, OA (sample 2); right, normal (sample 3). The top first study to investigate PCSCT for OA without the images include the femur, the tibia, and the patella. In need for any contrast agent. We have shown that the 60 the bottom images, the patella has been numerically re- keV and 70 keV monoenergetic images are optimal in moved to visualise the cartilage surface. Indeed, cartilage terms of noise and CNR, respectively, and that these surface defects and calcifications can be seen for the OA provide superior image quality compared to the HU specimen (left), with no defect apparent for the normal images. Fig. 3 Right: Energy fingerprints of the different joint tissues computed from the virtual monoenergetic images. Attenuation versus energy for the regions of interest shown in the left panel: patella cartilage (blue), soft tissue different from cartilage (red), cortical bone (yellow), and subchondral bone (green). Plot shows mean and standard deviation across subjects for the selected regions of interest Chappard et al. European Radiology Experimental (2022) 6:10 Page 6 of 10 Table 1 Clinical descriptions of the knee specimens in terms of cartilage, osteophytes, and bone cyts Sample Sex Age KL Lateral Medial Cartilage Femur Patella Femur Patella 1 F 89 4 Thin, defects Thin, defects Thin, defects Thin, defects calcifications+ 2 M 89 3 Thin, irregular Thin, irregular Thin, defects Thin, irregular calcifications+ calcifications+ calcifications+ calcifications+ 3 M 59 1 Normal, smooth Normal, smooth Normal, defects Normal, smooth 4 F 90 0 Normal, smooth Thin, defects Normal, smooth Normal, irregular calcifications+ 5 F 81 0 Normal, smooth Normal, smooth Normal, smooth Normal, smooth calcifications+ calcifications+ Calcifications+ 6 F 94 1 Normal, irregular Normal, irregular Normal, smooth Normal, smooth 7 M 63 2 Thin, defects Thin, defects Normal, defects Thin, defects Subchondral bone cysts 1 Large size, numerous Large size, numerous Large size numerous Large size, numerous 2 Middle size Small size Middle size – few numerous few 3 –– – – 4 Middle size Large size – Middle size numerous numerous few 5 –– – – 6 –– – – 7 Small size Small size –– few few Osteophytes 1 Large size Small size Large size Large size 2 – Small size Small size – 3 –– – – 4 –– – – 5 –– – – 6 –– – – 7 Small size Small size Small size Small size In a previous PCSCT study performed in vivo for the conventional detectors, which consequently decreases abdomen [18], 72 keV monoenergetic images showed the CNR. In contrast, for PCDs, the equally weighted en- improved contrast and SNR compared to conventional ergy photons have positive effects on the contrast [9]. images with similar attenuation patterns. In the present We hypothesise that the higher resolution and higher study, the monoenergetic images in the range from 60 CNR in soft tissue for PCSCT with respect to standard keV to 80 keV had lower noise than the conventional CT will lead to superior diagnosis potential for applica- images, with the 60 keV and 70 keV monoenergetic im- tions for which resolution and contrast are critical. In ages showing similar noise levels, although the lowest addition, another major advantage of virtual monoener- noise overall was for 60 keV. These data confirm the im- getic images is that they provide quantitative attenuation provements in the SNR demonstrated in a previous sim- measurements and reduce beam hardening artifacts [23]. ulated study that used a cadmium zinc telluride PCD in Finally, one of the major advantages of PCD detectors is comparison to conventional detectors [21]. Indeed, the the patient’s radiation dose reduction. Indeed, based on intrinsic qualities of PCDs can explain SNR and CNR in- an American College of Radiology accredited phantom, creases: first, the detector counts the number of pulses it was demonstrated that image quality can be main- greater than the preset threshold, which consequently tained with a reduction in the dose of 40 to 60% [24]. eliminates the electronic noise; secondly, high energy Different joint tissues showed specific energy finger- photons contribute more than low energy photons in prints. In particular, no overlap was seen between the Chappard et al. European Radiology Experimental (2022) 6:10 Page 7 of 10 Fig. 4 Transversal views of the selected 60 keV monoenergetic images for six specimens, the specimens 1 and 2 are severe osteoarthritis and sample 7 moderate osteoarthritis with numbers corresponding to samples. Red circle, cartilage calcifications; yellow circle, cartilage with different thicknesses; blue circle, cartilage defects; purple circle, bone cysts of various size; green circle: osteophytes of various size cartilage and the soft tissue surrounding the cartilage. energy monoenergetic images (i.e.,20−80 keV), it is pos- The analysis of the histograms showed that subjects with sible to differentiate different calcium deposits in periph- cartilage defects are shifted in the lower part of the his- eral joints, such as calcium pyrophosphate and calcium tograms and on the contrary the presence of calcifica- apatite [16, 17] In the present study, PCSCT allowed the tions shifted the histogram in the high part. characterisation of bone cysts, which are related to Indeed, it is possible to discriminate calcifications in mechanical stress in OA knees, and consequently this cartilage and the meniscus. CT has been proposed to will be interesting to study further for better understand- evaluate calcium deposition, which is especially useful in ing of the OA process [25] and it will also help to better the early stages of OA [24]. Using a preclinical scanner phenotype such OA patients [26] and cysts formation with high-resolution (i.e., 0.1 mm voxels) and low- following cartilage repair [27]. Indeed, bone cysts Fig. 5 Three-dimensional displays of the 60 keV monoenergetic for two selected samples. a, left, OA (sample 2); a, right normal (sample 3). White arrows show cartilage defects (top) and calcifications (bottom) on OA. b Illustration of segmented bone cysts (in red) with the semiautomatic region growing tool in OA sample (sample 2) Chappard et al. European Radiology Experimental (2022) 6:10 Page 8 of 10 Fig. 6 Synchrotron radiation images at 55 keV, PCSCT monoenergetic images at 60 keV and HR-pQCT images for 2 samples: a normal one (specimen 4); and an osteoarthritic one (specimen 7). The cartilage border with joint space is visible on the monoenergetic images based on synchrotron radiation, taken as references and the virtual PCSCT monoenergetic images. The yellow arrows correspond to cartilage defects Chappard et al. European Radiology Experimental (2022) 6:10 Page 9 of 10 Table 2 Quantitative analysis of the bone cysts for two of the these data suggest that quantitative measurements of specimens bone cysts in terms of numbers and size might lead Specimen Site Compartment Cysts to new biomarkers for better phenotyping of OA. The prototype is operational for clinical research, with fur- Number Volume (mm ) ther studies needed to validate these data on living Total Maximum patients. Sample 1 Femoral Medial 8 185.8 110.6 Lateral 44 355 106.1 Abbreviations CNR: Contrast-to-noise ratio; HU: Hounsfield unit; OA: Osteoarthritis; Total 52 540.8 110.6 PCD: Photon-counting detectors; PCSCT: Photon-counting spectral Patellar Medial 10 105.1 31 computed tomography; SD: Standard deviation; SNR: Signal-to-noise ratio Lateral 12 371 204.9 Authors’ contributions Total 22 476.1 204.9 All authors read, edited, and approved the final version of the manuscript. CC carried out the study concept, design, data acquisition, analysis and Sample 2 Femoral Medial 3 241.8 109.5 interpretation, and manuscript preparation. JA contributed to data Lateral 4 223.2 125.5 acquisition, data analysis, and manuscript preparation. CO contributed to data acquisition and analysis. SSM contributed to data acquisitions. LB Total 7 465 125.5 contributed to data analysis. JBP carried out the CT examination. PD contributed to study concept, design, interpretation. FP carried out the study Patellar Medial 4 13.2 5.5 concept, design, data acquisition, analysis, and manuscript editing. Lateral 11 37.9 17.7 Funding Total 15 51.1 17.7 This project was supported by the ANR project SALTO (ANR-17-CE19-0011- 01), within LabEx PRIMES (ANR-11-LABX-0063) of University de Lyon. It has received funding from the European Union Horizon 2020 Research and quantification were rarely performed from standard CT Innovation Program under Marie Sklodowska-Curie grant agreement N° 701915. The project also received funding from the European Union Horizon [26] and one time from high-resolution peripheral CT 2020 Research and Innovation Program under grant agreement N° 668142. [28]. This work was partly funded by France Life Imaging (grant ANR-11-INBS- The present study is subject to some limitations. 0006) from the French Investissements d’Avenir. The reconstructed images depend on the material de- Availability of data and materials composition and the postprocessing methods used. The datasets used and/or analysed during the current study are available Material decomposition was carried out on a photo- from the corresponding author on reasonable request. electric and Compton basis. Other specific material Declarations basis, such as those based on bone and soft tissue, or even more general basis, are possible [29]. The inves- Ethics approval and consent to participate tigation of other potentially improved basis might lead The study was approved by the Ethics Committee of Paris University, Paris. The tissue donors or their legal guardians provided informed written to further information with regard to the analysis of consent to give their tissue for investigations, in accord with legal clauses. cartilage and calcifications. The material decompos- itionmethodusedwas basedonconventionalmax- Consent for publication Not applicable imum likelihood, without regularisation in a pixel-by- pixel manner [11]. More advances inverse methods Competing interests including regularisation [30–33] or based on deep The authors of this manuscript declare no relationships with any companies whose products or services may be related to the subject matter of the learning approaches [34–36] might lead to improve- article. ments here. The image post-processing here was spe- cifically designed for knee images; it might be further Author details B3OA, CNRS UMR 7052, U 1271 Inserm, University of Paris, Paris, France. improved [37]. Our quantitative analysis is exploratory University of Lyon, INSA-Lyon, CNRS, Inserm, CREATIS UMR 5220, U1206, at this level, to show the potential of PCSCT for OA. Lyon, France. However, more refined methods in combination with Received: 27 September 2021 Accepted: 30 December 2021 a larger number of samples might lead to new bio- markers that translate into more accurate and meaning- ful OA scoring, and consequently to improved diagnosis. References Further experiments carried out in vivo and with more 1. Loeser RF, Goldring SR, Scanzello CR, Goldring MB (2012) Osteoarthritis: a disease of the joint as an organ. Arthritis Rheum 64:1697–1707. https://doi. subjects are needed to validate these findings. org/10.1002/art.34453 Epub 2012 Mar 5 In conclusion, virtual monoenergetic can provide 2. Thysen S, Luyten FP, Lories RJU (2015) Targets, models and challenges in direct visualisation of both cartilage and bone integ- osteoarthritis research. Dis Model Mech 8:17–30. https://doi.org/10.1242/ dmm.016881 rity, which are essential for the characterisation of 3. Berenbaum F (2010) Targeted therapies in osteoarthritis: a systematic review osteoarticular diseases that affect the whole joint, with of the trials on www.clinicaltrials.gov. Best Pract Res Clin Rheumatol 24:107– this demonstration thus focused on OA. In addition, 119. https://doi.org/10.1016/j.berh.2009.08.007 Chappard et al. European Radiology Experimental (2022) 6:10 Page 10 of 10 4. Magarelli N, Amelia R, Melillo N, Nasuto M, Cantatore F, Guglielmi G (2012) 24. Aslund M, Fredenberg E, Telman M, Danielsson M (2010) Detectors for the Imaging of chondrocalcinosis: calcium pyrophosphate dihydrate (CPPD) future of X-ray imaging. Radiat Prot Dosimetry 139:327–333. https://doi. crystal deposition disease -- imaging of common sites of involvement. Clin org/10.1093/rpd/ncq074 Exp Rheumatol 30:118–125 25. McErlain DD, Milner JS, Ivanov TG, Jencikova-Celerin L, Pollmann SI, 5. Burnett WD, Kontulainen SA, McLennan CE et al (2019) Knee osteoarthritis Holdsworth DW (2011) Subchondral cysts create increased intra-osseous patients with more subchondral cysts have altered tibial subchondral bone stress in early knee OA: a finite element analysis using simulated lesions. mineral density. BMC Musculoskelet Disord 20:14. https://doi.org/10.1186/ Bone 48:639–646. https://doi.org/10.1016/j.bone.2010.11.010 s12891-018-2388-9 26. Burnett WD, Kontulainen SA, McLennan CE, Hunter DJ, Wilson DR, Johnston JD (2014) Regional depth-specific subchondral bone density measures in 6. Metzger SC, Koehm M, Wichmann JL, et al (2016) Dual-energy CT in osteoarthritic and normal patellae: in vivo precision and preliminary patients with suspected gouty arthritis: effects on treatment regimen and comparisons. Osteoporos Int 25:1107–1114. https://doi.org/10.1007/s00198- clinical outcome. Acad Radiol 23:267–272. https://doi.org/10.1016/j.acra.201 013-2568-2 5.10.019 27. Gao L, Cucchiarini M, Madry H (2020) Cyst formation in the subchondral 7. Budzik JF, Marzin C, Legrand J, Norberciak L, Becce F, Pascart T (2021) Can bone following cartilage repair. Clin Transl Med 10:e248. Review. https://doi. dual-energy computed tomography be used to identify early calcium org/10.1002/ctm2.248 crystal deposition in the knees of patients with calcium pyrophosphate 28. Kroker A, Zhu Y, Manske SL, Barber R, Mohtadi N, Boyd SK (2017) deposition. Arthritis Rheumatol 73:687–692. https://doi.org/10.1002/art.41 Quantitative in vivo assessment of bone microarchitecture in the human knee using HR-pQCT. Bone 97:43–48. https://doi.org/10.1016/j.bone.2016.12. 8. Taguchi K, Iwanczyk JS (2013) Vision 20/20: Single photon counting X-ray detectors in medical imaging. Med Phys 40:100901. https://doi.org/10.111 29. Bornefalk H (2012) XCOM intrinsic dimensionality for low-Z elements at 8/1.4820371 diagnostic energies. Med Phys 39:654–657. https://doi.org/10.1118/1.36753 9. Si-Mohamed S, Bar-Ness D, Sigovan M, et al (2018) Multicolour imaging with spectral photon-counting CT: a phantom study. Eur Radiol Exp 2:34. 30. Ducros N, Abascal J, Sixou B, Rit S, Peyrin F (2017) Regularization of https://doi.org/10.1186/s41747-018-0063-4 nonlinear decomposition of spectral X-ray projection images. Med Phys 44: 10. Willemink MJ, Persson M, Pourmorteza A, Pelc NJ, Fleischmann D (2018) e174–e187. https://doi.org/10.1002/mp.12283 Photon-counting CT: technical principles and clinical prospects. Radiology 31. Kim K, Ye JC, Worstell W et al (2015) Sparse-view spectral CT reconstruction 289:293–312. https://doi.org/10.1148/radiol.2018172656 using spectral patch-based low-rank penalty. IEEE Trans Med Imaging 34: 11. Schlomka JP, Roessl E, Dorscheid R, et al (2008) Experimental feasibility of 748–760. https://doi.org/10.1109/TMI.2014.2380993 multi-energy photon-counting K-edge imaging in pre-clinical computed 32. Semerci O, Hao N, Kilmer ME, Miller EL (2014) Tensor-based formulation and tomography. Phys Med Biol 53:4031–4047. https://doi.org/10.1088/0031-91 nuclear norm regularization for multienergy computed tomography. Trans 55/53/15/002 Image Process IEEE 23:1678–1693. https://doi.org/10.1109/TIP.2014.2305840 12. Riederer I, Si-Mohamed S, Ehn S, et al (2019) Differentiation between blood 33. Touch M, Clark DP, Barber W, Badea CT (2016) A neural network-based and iodine in a bovine brain -- Initial experience with spectral photon- method for spectral distortion correction in photon counting X-ray CT. Phys counting computed tomography (SPCCT). PLoS One 14:e0212679. https:// Med Biol 61:6132–6153. https://doi.org/10.1088/0031-9155/61/16/6132 doi.org/10.1371/journal.pone.0212679 34. Arridge S, Maass P, Öktem O, Schönlieb C-B (2019) Solving inverse problems 13. Si-Mohamed S, Cormode DP, Bar-Ness D, et al (2017) Evaluation of spectral using data-driven models. Acta Numerica 28:1–174. https://doi.org/10.1017/ photon counting computed tomography K-edge imaging for determination S0962492919000059 of gold nanoparticle biodistribution in vivo. Nanoscale 9:18246–18257. 35. Bussod S, Abascal JFPJ, Ducros N et al (2019) In: Human knee phantom for https://doi.org/10.1039/c7nr01153a spectral CT: validation of a material decomposition algorithm (ed) 2019 IEEE 14. Si-Mohamed S, Tatard-Leitman V, Laugerette A, et al (2019) Spectral photon- 16th International Symposium on Biomedical Imaging (ISBI 2019), pp 1170– counting computed tomography (SPCCT): in-vivo single-acquisition multi- 1173. https://doi.org/10.1109/ISBI.2019.8759192 phase liver imaging with a dual contrast agent protocol. Sci Rep 9:8458. 36. Ravishankar S, Ye JC, Fessler JA (2020) Image reconstruction: from sparsity to https://doi.org/10.1038/s41598-019-44821-z data-adaptive methods and machine learning. Proc IEEE Inst Electr Electron 15. Ding H, Zhao B, Baturin P, Behroozi F, Molloi S (2014) Breast tissue Eng 108:86–109. https://doi.org/10.1109/JPROC.2019.2936204 decomposition with spectral distortion correction: A postmortem study. 37. Abascal JFPJ, Si-Mohamed S, Douek P, Chappard C, Peyrin F (2019) In: A Med Phys 41:101901. https://doi.org/10.1118/1.4894724 sparse and prior based method for 3D image denoising (ed) 2019 27th 16. Kirkbride TE, Raja AY, Müller K, Bateman CJ, Becce F, Anderson NG (2017) European Signal Processing Conference (EUSIPCO), pp 1–5. https://doi.org/1 Discrimination between calcium hydroxyapatite and calcium oxalate using 0.23919/EUSIPCO.2019.8902564 multienergy spectral photon-counting CT. AJR Am J Roentgenol 209:1088– 1092. https://doi.org/10.2214/AJR.17.18394 17. Stamp LK, Anderson NG, Becce F et al (2019) Clinical utility of multi-energy Publisher’sNote spectral photon-counting computed tomography in crystal arthritis. Arthritis Springer Nature remains neutral with regard to jurisdictional claims in Rheumatol 71:1158–1162. https://doi.org/10.1002/art.40848 published maps and institutional affiliations. 18. Große Hokamp N, Gilkeson R, Jordan MK, et al (2019) Virtual monoenergetic images from spectral detector CT as a surrogate for conventional CT images: Unaltered attenuation characteristics with reduced image noise. Eur J Radiol 117:49–55. https://doi.org/10.1016/j.ejrad.2019.05.019 19. Rajendran K, Löbker C, Schon BS, et al (2017) Quantitative imaging of excised osteoarthritic cartilage using spectral CT. Eur Radiol 27:384–392. https://doi.org/10.1007/s00330-016-4374-7 20. Roemer F, Eckstein F, Hayashi D, Guermazi A (2014) The role of imaging in osteoarthritis. Best Pract Res Clin Rheumatol 28:31–60. https://doi.org/10.101 6/j.berh.2014.02.002 21. Steadman R, Herrmann C, Livne A (2017) ChromAIX2: A large area, high count-rate energy-resolving photon counting ASIC for a Spectral CT Prototype. Nucl Instrum Methods Phys Res Sect Accel Spectrometers Detect Assoc Equip 862:18–24. https://doi.org/10.1016/j. nima.2017.05.010 22. Kellgren JH, Lawrence JS (1957) Radiological assessment of osteoarthritis. Ann Rheumat Dis 16:494–502. https://doi.org/10.1136/ard.16.4.494 23. Kalender WA (2014) Dose in X-ray computed tomography. Phys Med Biol 59:R129–R150. https://doi.org/10.1088/0031-9155/59/3/R129

Journal

European Radiology Experimental – Springer Journals

Published: Feb 22, 2022

Keywords: Bone cysts; Cartilage; Osteophyte; Osteoarthritis (knee); Tomography (X-ray computed)

There are no references for this article.