Hierarchical Multiscale Local Binary Pattern For Better Osteoporosis Detection

Currently, a number of clinical decision tools have been developed for osteoporosis risk assessment by measuring bone mineral density and/or analyzing bone images. Unfortunately, the great similarity and correlation between healthy and diseased bone images represents the major obstacle in building an effective osteoporosis detection system, making this task a major challenge. The aim of this work is to develop a texture analysis based osteoporosis detection system. Hierarchical Multiscale Local Binary Pattern (H-MLBP) technique is used in the feature extraction step. Thus, after selecting the dominant features, an emerging classifier, Ensemble Classifiers (EC), is used to classify them. The results obtained using a public database showed the robustness of the feature extraction method used to achieve high precision, which favors it to be effective for building an efficient osteoporosis detection system.