Automatic spinal cord segmentation from axial-view MRI slices using CNN with grayscale regularized active contour propagation

• Published in: Computers in biology and medicine Vol. 132; p. 104345
• Main Authors: Zhang, Xiaoran, Li, Yan, Liu, Yicun, Tang, Shu-Xia, Liu, Xiaoguang, Punithakumar, Kumaradevan, Shi, Dawei
• Format: Journal Article
• Language: English
• Published: United States Elsevier Ltd 01.05.2021; Elsevier Limited
• Subjects: 2D magnet resonance imaging; Algorithms; Automation; Central nervous system diseases; Cervical spondylotic myelopathy; Convolutional neural network; Deep learning; Image acquisition; Image processing; Image segmentation; Internal Medicine; Level set evolution; Magnetic resonance imaging; Metric space; Morphology; Neural networks; Other; Patients; Reliability analysis; Spinal cord; Spinal cord segmentation; Statistical analysis; Surgery; Trauma; 2D magnet resonance imaging; Level set evolution; Spinal cord segmentation; Convolutional neural network; Cervical spondylotic myelopathy
• ISSN: 0010-4825; 1879-0534; 1879-0534
• DOI: 10.1016/j.compbiomed.2021.104345

Accurate positioning of the responsible segment for patients with cervical spondylotic myelopathy (CSM) is clinically important not only to the surgery but also to reduce the incidence of surgical trauma and complications. Spinal cord segmentation is a crucial step in the positioning procedure. This study proposed a fully automated approach for spinal cord segmentation from 2D axial-view MRI slices of patients with CSM. The proposed method was trained and tested using clinical data from 20 CSM patients (359 images) acquired by the Peking University Third Hospital, with ground truth labeled by professional radiologists. The accuracy of the proposed method was evaluated using quantitative measures, the reliability metric as well as visual assessment. The proposed method yielded a Dice coefficient of 87.0%, Hausdorff distance of 9.7 mm, root-mean-square error of 5.9 mm. Higher conformance with ground truth was observed for the proposed method in comparison to the state-of-the-art algorithms. The results are also statistically significant with p-values calculated between state-of-the-art methods and the proposed methods. [Display omitted] •An integrated CNN and grayscale regularized active contour approach is proposed for the spinal cord segmentation.•The effectiveness of the proposed method is demonstrated through clinical data (20 subjects, 359 images).•The results indicate the feasibility of automated spinal cord segmentation based on axial-view 2D MRI slices.