Detection of distal radius fractures trained by a small set of X-ray\n images and Faster R-CNN

Abstract

Distal radius fractures are the most common fractures of the upper extremity\nin humans. As such, they account for a significant portion of the injuries that\npresent to emergency rooms and clinics throughout the world. We trained a\nFaster R-CNN, a machine vision neural network for object detection, to identify\nand locate distal radius fractures in anteroposterior X-ray images. We achieved\nan accuracy of 96\\% in identifying fractures and mean Average Precision, mAP,\nof 0.866. This is significantly more accurate than the detection achieved by\nphysicians and radiologists. These results were obtained by training the deep\nlearning network with only 38 original images of anteroposterior hands X-ray\nimages with fractures. This opens the possibility to detect with this type of\nneural network rare diseases or rare symptoms of common diseases , where only a\nsmall set of diagnosed X-ray images could be collected for each disease.\n

Ähnliche Arbeiten