Segmentation of Spinal Canal using Active Contour Model

Today, the lifestyle of people has paved the way for rise in spinal cord disorders. Severe cases have been reported which could be treated if diagnosed early. Scoliosis is a deformity to spine and ribs which is the primary cause of spinal curvature. The major challenge of scoliosis disease is the unnoticeable change in the orientation of spinal column at its early stage. Moreover, it is visually detectable only in the prodromal stage. The early diagnosis could help cure the disease through exercises and minor surgeries. Depending on manual diagnosis techniques is a tiring task as it can deliver inaccurate results. An automatic segmentation method which helps in the early diagnosis is proposed in this paper. Initially, CT image which is the input is fed into the system. CT images have high contrast between bone and adjacent tissues. Sagittal view datasets have been chosen in order to calculate the cob angle for the measurement of scoliosis intensity. Further, distortions are removed from the image and pre processing is performed followed by K-means clustering which detects the spinal canal. In order to segment the required features, the output of clustering is loaded to Active Contour Model. Finally, segmentation of spinal canal is completed. Experimental results prove the accuracy of 95%,86.86%,92.22% for Lumbar Vertebrae CT , Lumbar spine CT, Lumbar spine CT with multiple compression fractures respectively for the proposed system which is greater than traditional diagnosis methods. Subsequently, this would be a revolutionary study which assists the doctors for the early diagnosis of the disease.