Lumbar Spine Classification using Pyramidal Histogram of Oriented Gradients

Objective: To diagnose any injury in the lumbar region of human spine, the classification of each vertebra and Intervertebral Disc (IVD) is the vital task. Methods: The classification of the lumbar structure is done using Pyramidal Histogram of Gradients (PHOG). The PHOG technique is applied on the MR Image to select the basic intensity level features. These features are trained with Support Vector Machine (SVM). After building a model with SVM classification algorithm, classify the MR Images for discs and vertebrae separately. Findings: The accuracy is calculated to verify the performance of this work. This classification procedure is performed on a lumbar MR image dataset which contains 960 IVDs and 800 vertebrae with T1 and T2 weighted for 80 subjects.