Deep Neural Network for the Automatic Classification of Vertebral Column Disorders

Abstract

In the human body, the vertebral column consists of vertebras, nerves, invertebrate discs, medulla, joints, and muscles, which provides support for body and movement axle. Dysfunction to any of the above components in this complex system creates disorders like Disc hernia and Spondylolisthesis. Manually classifying these disorders is a difficult task. Recently Machine learning (ML) techniques were applied in automating the vertebral column disorder classification. In this work, we applied Deep Neural Network (DNN) to classify the vertebral column dataset with three classes (Normal, Disk Hernia, and Spondylolisthesis). The vertebral column dataset was collected from the UCI machine learning database repository and has 310 records for training and testing with six biomechanical attributes. The classification accuracy and F-score for the DNN classifier in the vertebral column dataset is 85% and 83% respectively. Comparison with existing ML systems shows that our DNN based classification approach exhibits promising results.