Misshapen Pelvis Landmark Detection by Spatial Local Correlation Mining for Diagnosing Developmental Dysplasia of the Hip

Abstract

Developmental dysplasia of the hip (DDH) refers to an abnormal development of the hip joint in infants. Accurately detecting and identifying the pelvis landmarks is a crucial step in the diagnosis of DDH. Due to the temporal diversity and pathological deformity, it is a difficult task to detect the misshapen landmark and diagnose the DDH illness condition for both human expert and computer. Moreover, there is no adequate and public dataset of DDH for research. In this paper, we investigate the spatial local correlation with convolutional neural network (CNN) for misshapen landmark detection. First, we convert the detection of a landmark to the detection of the landmark’s local neighborhood patch, which yields effective spatial local correlation for the identification of a landmark. Then, a deep learning based method named FR-DDH network, is proposed for misshapen pelvis landmark detection. It mines the spatial local correlation and detects the best-matched region according to the spatial local correlation. To the end, the landmarks are located at the center of the regions. Besides, a dataset with 9813 pelvis X-ray images is constructed for research in this area, and it will be released for public research. To the best of our knowledge, this is the first attempt to apply deep learning in the diagnosis of DDH. Experimental results show that our approach achieves an excellent precision in landmark location (MAE 1.24 mm) and illness diagnosis over human experts.