AUTOMATIC SEGMENTATION OF CERVICAL VERTEBRAE USING MACHINE LEARNING APPROACH

Abstract

The first seven vertebrae, of the spine, build the cervical, spine which supports, the head's weight, protects and surrounds the spinal cord, and allows for a wide variety of head motions. Due to the flexible nature of the Cervical spine it is quite vulnerable to injuries. This section of the spine is affected by a variety of disorders, includes poor posture in which head is forward and out in front of your shoulders this posture puts excess stress on the vertebrae in your cervical spine. Unfortunately, due to human error, a huge majority of injuries in cervical X-ray images go undetected and incorrect diagnosis may lead to serious long-term consequences. As previous research shows abnormality in cervical structure may affect the movement of head. The foundation for spinal image analysis is precise vertebral segmentation. A totally automatic, segmentation, for cervical, vertebrae,in X-ray pictures that assists and improves clinical interpretation due to the success of the computer-aided systems in numerous medical imaging is suggested in this dissertation. For accurate segmentation object detection algorithm, precisely YOLO v5 has been used. Without any human involvement, the framework can, take an X-ray, image, and deliver a cervical vertebral segmentation. Each component of completely automated architecture was trained on 377 X-ray pictures and then tested on additional 120 individuals dataset, which is recorded in special circumstances. YOLO v5 is used to train the model, segmentation of cervical vertebrae is performed and each cervical vertebrae is labelled using YOLO v5 in this way it can assist physician and doctors in their work.