Fully automatic cervical vertebrae segmentation framework for X-ray images

• Published in: Computer methods and programs in biomedicine 2018-04, Vol.157, p.95-111
• Main Authors: Al Arif, S. M. Masudur Rahman, Knapp, Karen, Slabaugh, Greg
• Format: Article
• Language: English
• Subjects: Algorithms; Automation; Cervical vertebrae; Cervical Vertebrae - diagnostic imaging; Deep learning; Diagnostic Errors; FCN; Humans; Image Interpretation, Computer-Assisted - methods; Localization; Neural Networks (Computer); Probability; Radiography - methods; Segmentation; UNet; X-ray
• ISSN: 0169-2607; 1872-7565
• DOI: 10.1016/j.cmpb.2018.01.006

•A deep segmentation network based spine localization algorithm which outperforms the previous state-of-the-art by a large margin.•A novel spatial probability prediction deep convolutional network which achieves human-level performance in localizing vertebrae centers.•A novel shape-aware deep segmentation network for vertebrae segmentation.•A first of its kind fully automatic framework which combines the global localization, center localization and vertebrae segmentation in a single thread and provides a segmentation result for a real-life emergency room X-ray images without any manual input. The cervical spine is a highly flexible anatomy and therefore vulnerable to injuries. Unfortunately, a large number of injuries in lateral cervical X-ray images remain undiagnosed due to human errors. Computer-aided injury detection has the potential to reduce the risk of misdiagnosis. Towards building an automatic injury detection system, in this paper, we propose a deep learning-based fully automatic framework for segmentation of cervical vertebrae in X-ray images. The framework first localizes the spinal region in the image using a deep fully convolutional neural network. Then vertebra centers are localized using a novel deep probabilistic spatial regression network. Finally, a novel shape-aware deep segmentation network is used to segment the vertebrae in the image. The framework can take an X-ray image and produce a vertebrae segmentation result without any manual intervention. Each block of the fully automatic framework has been trained on a set of 124 X-ray images and tested on another 172 images, all collected from real-life hospital emergency rooms. A Dice similarity coefficient of 0.84 and a shape error of 1.69 mm have been achieved.