Evaluation of Local Binary Pattern for Osteoporosis Classification

Abstract

Deterioration of the bone’s microarchitecture and low bone mineral density, which results in increased fragility of bone, are symptoms of the disease osteoporosis, which decreases bone mass. Early osteoporosis identification can prevent the disease and predict fracture risk. Usually, the diagnosis is based on the analysis of X-ray images. However, the healthy and osteoporotic subject radiography shows a great resemblance. This study aims to develop an evaluation of an automatic osteoporosis identification system based on texture analysis. This paper proposes a Local Optimal Oriented Pattern (LOOP) to address some of the shortcomings of existing feature descriptors such as Local Binary Pattern (LBP) and Local Directional Pattern (LDP). Ensemble and SVM learning algorithms were used for the classification task. The obtained results were compared with some state-of-art methods used in the literature. Experimental results show that the proposed approach outperforms the previous binary descriptor in terms of recognition accuracy proving that the proposed approach is efficient for real clinical applications.