Detection of Upper Limbs Fractures from X-Ray Images Using Machine Learning Techniques

Abstract

Bone fractures are common injuries which might affect individuals at any age. The diagnosis should be accurate for proper treatment. Although conventional X-ray imaging is generally practiced, physicians might not be able to diagnose the fractures due several challenges such as infinitesimal fractures that could lead to the misdiagnosis. The misdiagnosis is probably happened especially in the case of the small size of bones. This research is carried out to improve detection of upper limbs fractures. In each upper limbs there are 32 bones. 27 bones are in balms which are quite small pieces. Therefore, an accurate diagnosis is really requires. Automatic diagnosis of bone fractures has already been exploited. However, there are several issues due to the accuracy rate. The proposed framework in this study is an experiment to obtain high level of results. It involves: data gathering, preparation, feature extraction, selection of the best features, and classification. It employs advanced methods such as Gray Level Co-occurrence Matrix, Local Binary Patterns, and deep learning models like VGG16 and DenseNet to improve the diagnosis accuracy. The findings have shown that the SVM classifier yields a rate accuracy of 96%, which outperforms other methods. These findings have been yielded from analyzing the dataset of 8942 samples. This result shows that the proposed framework has outperform other studies.

Ähnliche Arbeiten