Classification of Intervertebral Disc on Lumbar MR Images using SVM

Computer aided diagnosis is the most predominant research in computer vision. While considering the lumbar Magnetic Resonance Images (MRIs), identification of pathologies is a complex task as the size of the soft tissues may vary. Before recognizing the injury in the soft tissue, it is necessary to identify in which soft tissue, the pathology occurs. In order to address this issue, the classification of lumbar Intervertebral Discs (IVDs) on lumbar MRI is performed. The classification is determined after feature extraction on the region of interest. Histogram of Oriented Gradients (HOG) is applied to extract the features and a model is built using Support Vector Machine (SVM) for classification. A multi-class classification is performed on 80 subjects, both T1-weighted and T2-weighted, with a total of 960 Intervertebral Discs. The result of the research work done in this paper gives 71.53% accuracy on the classification of Lumbar MR Images. Also, the classification accuracy is compared with the features extracted using Fourier Descriptors (FD), and it shows HOG gives a promising result.