Surgical Design Optimization of Proximal Junctional Kyphosis
Surgical Design Optimization of Proximal Junctional Kyphosis
Peng, Li;Zhang, Guangming;Zuo, Heng;Lan, Lan;Zhou, Xiaobo
2020-09-21 00:00:00
Hindawi Journal of Healthcare Engineering Volume 2020, Article ID 8886599, 8 pages https://doi.org/10.1155/2020/8886599 Research Article 1 1 1,2 1 3 Li Peng, Guangming Zhang, Heng Zuo, Lan Lan , and Xiaobo Zhou West China Biomedical Big Data Center, West China Hospital/West China School of Medicine, Sichuan University, Chengdu 610041, China School of Mathematics, Sichuan Normal University, Chengdu 610066, China Center for Computational Systems Medicine, School of Biomedical Informatics, University of Texas Health Science Center at Houston, Houston 77030, USA Correspondence should be addressed to Lan Lan; lanl@scu.edu.cn Received 7 July 2020; Revised 19 August 2020; Accepted 1 September 2020; Published 21 September 2020 Academic Editor: Antonio Gloria Copyright © 2020 Li Peng et al. 'is is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. Purpose. 'e objective of this study was to construct a procedural planning tool to optimize the proximal junction angle (PJA) to prevent postoperative proximal junctional kyphosis (PJK) for each scoliosis patient. Methods. Twelve patients (9 patients without PJK and 3 patients with PJK) who have been followed up for at least 2 years after surgery were included. After calculating the loading force on the cephalad intervertebral disc of upper instrumented vertebra of each patient, the finite-element method (FEM) was performed to calculate the stress of each element. 'e stress information was summarized into the difference value before and after operation in different regions of interest. A two-layer fully connected neural network method was applied to model the relationship between the stress information and the risk of PJK. Leave-one-out cross-validation and sensitivity analysis were implemented to assess the accuracy and stability of the trained model. 'e optimal PJA was predicted based on the learned model by optimization algorithm. Results. 'e mean prediction accuracy was 83.3% for all these cases, and the area under the curve (AUC) of prediction was 0.889. And the output variance of this model was less than 5% when the important factor values were perturbed in a range of 5%. Conclusion. Our approach integrated biomechanics and machine learning to support the surgical decision. For a new individual, the risk of PJK and optimal PJA can be simultaneously predicted based on the learned model. such as the increase of the proximal junction angle (PJA, the 1. Introduction sagittal Cobb angle between the inferior endplate of the For adolescent idiopathic scoliosis (AIS) patients, ortho- upper instrumented vertebra (UIV) and the superior end- pedic operations are employed to reconstruct the coronal plate of two cephalad vertebrae ) [6]. Proximal junctional and sagittal alignment to maintain the stability of spine [1]. kyphosis (PJK), an abnormal kyphotic deformity involving Long posterior instrumentation and fusion surgery is often a spinal segments proximally adjacent to the fusion segments, powerful surgical treatment for spinal deformity [2, 3]. has drawn the attention of many spine surgeons [7–9]. 'is frequent complication might cause regional pain, diminish During the treatment, vertebrae are fused using pedicle screws or other combinations of devices. Such fusion quality of life, and ultimately lead to revision surgery in some treatment is intended to reconstruct spinal geometry by severe cases [10, 11]. 'e generally accepted definition of PJK strong correction and derotation of the spine [4]. However, is described by Glattes et al. [8] that PJA is more than 10 and the lack of mobility in fusion segments has raised a pos- at least 10 greater than the preoperative measurement. 'e tulation that such fusion may increase the stress in proxi- incidence rate of PJK ranges extensively from 6.0% to 45.1% mally cephalad spinal segments and eventually accelerate [6, 8, 10, 12–14]. A retrospective review of 836 adult cases deterioration of the neighboring discs [5]. reported a higher percentage of unplanned readmission due Powerful correction maneuvers of predominantly all- to PJK within 90 days from surgery (51.9%) compared with pedicle instrumentation could also result in a series of issues other surgical complications [15]. 2 Journal of Healthcare Engineering 'e recent advances in computer-aided design (CAD) Training Testing phase have been rapidly changing the landscape in scoliosis phase treatment procedures, improving the clinical outcomes of A new patient patients significantly as valuable models are practiced [5, 16]. Preoperative Preoperative Many studies biomechanically assess and evaluate the in- Age Gender CT data CT data dependent effects of different instrumentation variables by Pre- and postoperative PJA Preoperative PJA from the finite-element method (FEM) [15, 17–21]. 'e difficulty from X-rays X-rays mainly comes from a finite range of optimizations by manual Simulation of a PJA on intervertebral disc tissue selection. Our study focuses on designing a reliable auto- matic system to assist surgeons designing and preoperatively help decrease the readmission rate of scoliosis patients. Stress calculated by FE Stress calculated by FE 'e larger difference value of preoperative and postop- Y: whether PJK erative PJA (more than 5 ) has been determined as a risk Optimization methods occurred or not factor of PJK [8]. However, the individual and coupling biomechanical effects of PJA are not yet fully understood. 'e Machine learning model purpose of this paper is to predict which AIS patients have The optimal training between Y, stress postoperative PJA higher risks of PJK due to biomechanical factors and calculate features, age, and gender the optimal PJA for each AIS patient. In this work, we hy- Figure 1: 'e flowchart of optimizing the surgical design of PJK. In pothesize that inappropriate intraoperative PJA change the training phase, the machine learning model is generated. compared with preoperation may lead to inhomogeneous During the testing phase, the model can be used for PJA opti- distribution of loading and ultimately result in various de- mization based on a new patient’s information. grees of degeneration in the cephalad intervertebral disc of UIV, and a suitable angle is beneficial for an individual patient to have a better prognosis after corrective surgery. 'e bio- 1 mm slice intervals and a 512 × 512 acquisition matrix, and then imported into Mimics 20.0 Imaging Software (Mate- mechanical information of intervertebral discs in the proxi- mal junctional segment following spinal deformity surgery rialise, Leuven, Belgium) for segmenting as shown in can be accurately simulated by integrating a finite-element Figure 2. Initial segmentation was performed by thresh- method (FEM) with a statistical learning model. olding image from 50 to 150 Hounsfield units chosen to most accurately preserve intervertebral disc geometry. Manual segmentation was employed to delineate regions, 2. Materials and Methods which were visible but could not be captured by automated 'is study presented an integrated approach to accurately methods. To limit the area-of-interest and reduce compu- simulate cephalad intervertebral disc behavior of the UIV for tational complexity, we restricted the zone to the cephalad pre- and postoperation, respectively, for the purpose of intervertebral disc of UIV because no tissue deformations optimizing the PJA for AIS patient. Figure 1 describes the appeared in adjacent vertebrae (as shown in Figure 2(b)). flowchart of the whole process. For spine corrective operation, a preprocedural plan is 'is retrospective study was completed with AIS patients meaningful only if it can be accurately transferred to a patient at who undergone scoliosis correction surgery from 2013 to the time of intervention. For this reason, we applied a validated 2018 at West China Hospital and have been followed-up for software (Surgimap, version 2.2.15.5), which could measure the at least 2 years to assess whether or not developing PJK. degree of the curvature quickly on pre- and postprocedural Patients were excluded if imaging information including standing radiographs, thus determining the influence of gravity preoperative, immediate postoperative (3–7 days after sur- at each vertebral level of patients with scoliosis in the upright gery) X-ray, and preoperative CT could not be obtained. position [22]. 'e severity of scoliosis can be evaluated by Twelve cases were recruited for this biomechanical study. measuring the Cobb angle. With using strong correction of all- For each patient, the collected data involved the upper pedicle instrumentation, the PJA changes with reconstructing instrumented vertebral documentation, actual PJA, gender, the coronal and sagittal alignment [6]. After selecting the most age, and preoperative spinal computed tomography (CT). suitable UIV according to the scoliosis type, assuming different Based on the postoperative 2-year PJA, which is more than stress distribution of the UIV surface would derive from an- ° ° 10 and at least 10 greater than the preoperative mea- gular variations of PJA, and inappropriate intraoperative PJA surement, those patients were categorized into 2 groups: PJK change compared with preoperation could lead to inhomo- group and non-PJK group. In this study, we used the geneous stress distribution of the upper body weight on the aforementioned study information for each patient as UIV surface and the degeneration of adjacent intervertebral ground truth to confirm the patient selection. disc. Usually, a virtual proximal segment correction will de- crease the PJA as possible. However, due to the biomechanical properties of tissues, patients may still be at risk for PJK two 2.1. Feature Extraction years after spinal surgery because of the nonspecific PJA. 2.1.1. Intervertebral Disc Segmentation and Quantification. 'erefore, our approach addresses the need to develop a re- liable process for simulating tissue behavior changes of in- Twelve subject-specific geometries of patients were acquired using preoperative CT scans. All images were acquired with tervertebral disc between pre- and postprocedure. Journal of Healthcare Engineering 3 Preoperative CT Intervertebral disc mesh data Finite element analysis Eight subregions FPL FPR Preoperative or NPR NPL postoperative X-ray Calculate loading force NAL NAR FAR FAL (a) (b) (c) (d) Figure 2: Illustration of the biomechanical analysis process. (a) Blue rectangle means the upper instrumented vertebra (UIV), the red one is UIV + 1, and the green one is UIV + 2. (b) In the mesh data of segmented cephalad intervertebral disc of UIV, light red indicates nucleus, and dark red demonstrates anulus fibrosus; white solid lines in an enlarger X-ray represent the inferior endplate of UIV and the superior endplate of two cephalad vertebrae, respectively, and the angle formed by the intersection of them is the proximal junctional angle (PJA, α). (c) Loading force calculated in the previous stage on the segmented vertebra by Altair OptiStruct. (d) F, anulus fibrosus; N, nucleus; A, anterior; P, posterior; R, right; L, left. 2.1.2. Loading Force Calculation with the PJA. To study the for these two components were given by the previous work depending on biomechanical analysis and the material mechanical behavior of the proximal intervertebral disc, we calculated the loading force on the tissue before and after the model as shown in Table 1 [5, 28]. 'ese parameters were surgical correction. We denote α as the PJA at pre- or used to simulate tissue biomechanical behavior based on postoperation. 'e loading force F (i.e., contact force) is Hooke’s law. And the heterogeneous properties of the in- perpendicular to the upper surface of the intervertebral disc tervertebral disc tissue will be examined in the proposed tissue, which is roughly equivalent to the decomposing force studies. of the patient’s gravity at this point. 'e decomposition angle of gravity can be estimated as α. Hence, F can be calculated by G∗ cosα, where G denotes the gravity of the 2.2. Stress Formulation. We extracted stresses as one of the biomechanical characteristics from FEM. 'e stress for each body weight above UIV (as shown in Figure 2(b)) [14]. mesh node varies according to different components of the 'e 3D segmented intervertebral discs were first dis- intervertebral disc. To obtain a distribution of stress features, cretized into small mesh by HyperMesh (Altair, USA). To we first simulated the intervertebral disc behavior obtain high precision mesh models, they were composed of tetrahedral elements, and each element contained 4 mesh responding to loading force. Denote stress features as nodes tetrahedral. 'e mesh nodes could be classified into � σ(G , α , E, v), where G denotes the force of gravity i i i i the boundary and free nodes; meanwhile, the boundary from the body weight above UIV of the ith patient, and α is the pre- or postoperative PJA. Next, the stress value σ is nodes were located in the inferior surface, which would be fixed in all degrees of freedom. We restricted the zone to the employed as the biomechanical feature of the ith patient. Finally, σ(·) represents the stress modeled by Hooke. Here, adjacent intervertebral disc of the UIV for focusing on the deformable area of interests with regard to PJK. FEM was implemented in the commercial software Altair OptiStruct. To validate our model, we loaded the pressure difference of pre- and postoperative force of one patient on 2.1.3. Assignment of the Intervertebral Disc Properties. his FEM model and compared the simulation results with Intervertebral disc (IVD) tissue is composed of a nucleus the true geometric model of the intervertebral disc generated pulposus (translucent gel) and an anulus fibrosus (lamellar from postoperative CT; the absolute error of volume was structure), with negligible vascularization in the anulus and 500 mm , while the relative error was 2.5%, and the absolute nucleus regions [23, 24]. Nonoriented collagen fibrils en- error of average disc height on the central sagittal plane was mesh in the proteoglycan-water pulposus and are sur- 0.3 mm, while the relative error was 5.0%, indicating that our rounded by the anulus fibrosus, a series of concentric simplified model could save labor and machine time based encircling lamellae with two well-defined axes of orientation on not influencing the authenticity of the FEM model [17]. [23, 25–27]. To simplify the analysis process, we defined the To elaborate the stress variations of corresponding re- intervertebral disc tissue as a linear elastic tissue with the gions caused by the selection of PJA, stress information was homogenous and isotropic properties [5]. Different material evaluated on eight anatomical regions of anulus fibrous and parameters in terms of Young’s modulus and Poisson’s ratio nucleus: left anterior, left posterior, right anterior, and right 4 Journal of Healthcare Engineering Table 1: Material parameters of the intervertebral disc tissue fully connected network was employed to efficiently model [5, 28]. nonlinear functions with less parameters [18, 19]. 'e clinical outcome y /whether PJK occurred or not of ith Young’s modulus (MPa) Poisson’s ratio patient can be modeled as the following equation: Nucleus 1.0 0.49 1 2 Anulus fibrosus 3.4 0.45 y � gΔσ