Ensemble learning-based abnormality diagnosis in wrist skeleton radiographs using densenet variants voting

10.48129/kjs.splml.19477

Abstract

Almost one out of the five people, including children, suffers from musculoskeletal disorders. It is the second biggest cause of disability. It affects the musculoskeletal system's major area, i.e., shoulder, forearm, and wrist, causing severe pain, joint noises, and disability. To detect the abnormality, radiologists analyze XRAYs taken from the different orientations of the patient's body. The automatic diagnosis of abnormality in the musculoskeletal system is a challenging task. In the past, many researchers detected the abnormality in the musculoskeletal system from radiograph images by using several deep learning techniques. They used capsule network, 169-layer convolutional neural network, and group normalized convolutional neural network. However, more work needs to be done to propose methods with improvements as the accuracy for the conventional methods is far away from 90$\%$. This paper presents an ensemble learning-based classification system for detecting abnormality in wrist radiographs. Tags in radiographs may result in learning noisy features hence reducing the performance. Therefore, tags are segmented and removed using UNet trained on the annotated ground truths. Segmented images are then used for voting-based diagnosis. The simulation results show that the proposed methodology improves testing accuracy by 1.5-4.5% compared to the available wrist skeleton abnormality diagnosis methods. The proposed methodology may be adopted to diagnose any kind of musculoskeletal abnormality.