Texture analysis using dual tree m-band and Rényi entropy. Application to osteoporosis diagnosis on bone radiographs
Résumé
In this paper, we propose a new method based on texture analysis for trabecular bone disease diagnosis. For this purpose, we present a working model based on a preprocessing step followed by a projection on a 1D oriented axis at a defined angle. First, the dual-tree transform M-band is applied on the 1D obtained signal. Then, the Rényi based information measure is computed on sub-bands coefficients in order to characterize the anisotropy which is strongly present in our application related to bone radiograph characterization. Finally, the Wilcoxon test is used on the Rényi entropies obtained for each subband and the K nearest neighborhoods classifier is used with the Rényi divergence as a distance. Applied on two different populations composed of osteoporotic (OP) patients and control (CT) subjects, a classification rate of 98% is achieved which provides a good discrimination between OP patients and CT subjects.