DETECTION OF FEMUR AND RADIUS FRACTURES IN X-RAY IMAGES
Sher Ee Lim, Yage Xing, Ying Chen, Wee Kheng Leow, Tet Sen Howe, Meng Ai Png · 2004
13 % of men and 30%–40 % of women suffer from osteoporotic bone fractures worldwide. In large hospitals, doctors need to visually inspect a large number of x-ray images to identify the fracture cases, which typically constitute a small fraction of all the x-ray images examined. After looking through many images containing healthy bones, a tired radiologist has been found to miss a fractured case among the many healthy ones. Automated fracture detection can help the doctors by screening for obvious cases and flagging suspicious cases for closer examinations. Since bone fractures can occur in many ways, no one single algorithm can detect all the possible fractures accurately. This paper describes an approach in detecting fractures of the femur and the radius by combining various detection methods. These methods extract different kinds of features for fracture detection. They include neck-shaft angle, which is specifically extracted for femur fracture detection, and Gabor texture, Markov Random Field texture, and intensity gradient, which are general features that can be applied to detecting fractures of various bones. Two types of classifiers are tested, namely, Bayesian classifier and Support Vector Machine. Test results show that the combined approach can improve both the fracture detection rate and the classification accuracy significantly compared to any single method.