Parametric Evaluation of Improved Deep Learning Networks for Musculoskeletal Disorder (MSD) Classification

Abstract

Over the past few decades, the major debate regarding healthcare throughout the world is the analysis, and findings of diseases by investigating the medical images. Musculoskeletal disorder classification from a massive radiological image archive has always been a tedious task for radiologists. In recent literature, deep learning paves its way towards biomedical image classification with maximum accuracy and efficiency. Besides, deep learning models have already outperformed in various medical applications. Specifically, Convolution Neural Network (CNN) and LSTM architecture have been widely used. In this paper, new variants of conventional deep learning models have been proposed. Subsequently, an exhaustive parametric comparison from the existing pre-trained model has been established to validate the improved efficacy and productivity.