Texture Aware Deep Features for Precise Wrist Fracture Detection

Bone fractures are a critical medical condition requiring a thorough and accurate diagnosis to ensure proper treatment and healing. The physical examination of X-ray images by radiologists is time-consuming and subject to human error, indicating the need for an automated solution. This study proposes an advanced deep-learning approach to detect and classify bone fractures from X-ray images. This research aims to improve detection accuracy and streamline the diagnostic process using cutting-edge computer vision methods. Our approach starts with preprocessing the dataset, encompassing image resizing to 50x50 pixels, enhancement, and augmentation to improve diversity. Canny Edge detection is used to emphasize structural edges in the augmented dataset. The dataset is classified into training, testing, and validation subsets. A Convolutional Neural Network (CNN) automatically extracts deep learning features, obtaining complex patterns that show bone fractures. The Gray Level Co-occurrence Matrix (G.L.C.M) is utilized for texture features. These features are arranged to form a comprehensive feature set. The principal component analysis (PCA) is applied to this fused feature set to shrink dimensionality while preserving critical information, which is then used for fracture prediction. Experiments were conducted using a dataset of 14,718 X-ray images covering various bone fractures. The proposed method achieved an incredible classification accuracy of 98%, significantly outperforming traditional diagnostic methods and other contemporary models.