Income inequality based texture feature extraction for human bone fracture detection and classification using machine learning

Abstract

Computer vision, machine learning, and digital image processing are very popular for early diagnosis of medical images. Bone fracture is growing day by day due to accidents or bone cancer and doctors use X-ray or CT scanning images to detect bone fracture. Therefore, it is necessary to develop a tool that will help the doctors for bone fracture analysis. The proposed research is capable enough to classify the bone as fracture and healthy. Initially, a 3x3 median filter is applied to remove the noise from the captured image then the Canny edge detection algorithm applied to locate edges and Hough transforming method, based on the matrix of the income inequality is used to identify the fracture region. The training and classification is done through Support Vector Machine (SVM). In the past similar work has been conducted but their research is focused only on the fracture bone classification. The data set of the previous works is based on a single type of bone. The proposed research can classify the healthy and the fractured bone. The present work data set is a collection of different human bones, which is unique compared to past researches. The proposed approach classification accuracy of the fractured bone is more than 95%, which is much better than [6] of 94.43% and [24] of 85.