Improving the Classification of Scoliosis on Radiographic Image using the AdaBoost Ensemble Model

Abstract

Scoliosis is a disorder in which the spine bends to one side or the other. Surgeons, physiatrists, and academicians can be confused when facing situations involving certain types of scoliosis that resemble the normal spine. Manually detecting scoliosis requires a lot of time and effort. The need for a method that can speed up the process by using an approach that surgeons, physiatrists, and academicians understand would undoubtedly solve the issues. To overcome this issue, a machine learning using image processing is introduced. A Grey-Level Co-Occurrence Matrix (GLCM) was implemented with an ensemble classification, which is AdaBoost, to classify between normal and scoliosis radiographic images. Based on the results, this method achieved an accuracy of 86.67%. The study would aid in the identification of more types of scoliosis by providing more data for future studies. Furthermore, it is hoped that it would be able to assist orthopaedics in making decisions.