Identification and Classification of Scoliosis Using CNN And SVM Algorithms

Abstract

Scoliosis is one of the most common diseases that is thrown under the radar. The  lateral curvature and rotation seen in the vertebrae of the spinal column is classified in mostly 2 types: C-Curve and S-Curve. People need to be aware of the early signs and symptoms so that it is diagnosed in the proper time and manner.  Scoliosis is mostly diagnosed and  identified by taking X-ray medical images of the spine and on the basis of the sideways curvature image modality. In traditional scoliosis diagnosis detection, the treatment is based on spinal assessment which is a manual study consisting of major limitations mainly being very tedious, time-consuming and cost effective. Till date, without the assistance of technology, scoliosis diagnosis has been a critical task in the beginning because it all depended on the patient history records or even the captured x-ray images of the patient. Hence to ease up the process and provide better treatment experience, our research work is focused on categorizing the type of scoliosis in an effective, accurate and reliable way. This is obtained by analyzing and processing  the input X-ray images obtained from the datasets and even the patients suffering from scoliosis. To overcome the aforementioned limitations, we developed a point-based automated method at different regions of the spinal cord resulting in accurate results using the Convolutional Neural Network (CNN) algorithm and further compared it to Support Vector Machines (SVM) for better understanding