Abstract

Bones are a vital component of the human body. Bone provides the capacity to move the body. Humans have a high rate of bone fractures. The X-ray image is used by the doctors to identify the fractured bone. The manual fracture identification technique takes a long time and has a high risk of mistake. Machine learning and artificial intelligence are critical in resolving difficult difficulties in clinical imaging. Both medical practitioners and patients benefit from machine learning and artificial intelligence. Nowadays, an automatic system is built to detect abnormalities in bone X-ray pictures with great accuracy. To achieve high accuracy with limited resources, image pre-processing methods are employed to improve the quality of medical images. Image preprocessing entails steps such as noise removal and contrast enhancement, resulting in an instantaneous abnormality detection system. In image classification challenges, the Gray Level Cooccurrence Matrix (GLCM) texture features are commonly utilised. The second order statistical information about grey levels between nearby pixels in an image is represented by GLCM. In this work, we used various machine learning algorithms to categorise the MURA (musculoskeletal radiographs) dataset’s bone X-ray images into fractures and no fracture categories. For anomaly detection, the four different classifiers SVM (support vector machine), Random Forest, Logistic Regression, and Decision tree are utilised. The aforementioned abnormality detection in X-ray pictures is evaluated using five statistical criteria, including Sensitivity, Specificity, Precision, Accuracy, and F1 Score, all of which indicate considerable improvement.

Authors

A. Selin Vironicka and J.G.R. Sathiaseelan
Bishop Heber College, India

Keywords

Machine Learning, GLCM, SVM, Random Forest, Logistic Regression, Decision Tree, MURA, Bone Fractures