Fractal analysis to BEMD’s IMFs: Application CT-Scan

Abstract

Osteoporosis is a serious disease due to the fractures it causes which can lead to pain, impotence, loss of independence and excess mortality (femoral neck fractures). In addition, it is a disease with a high recurrence rate and is age-specific. Therefore, the impact of osteoporosis on the already sensitive healthcare system will increase, and thus several preventive measures can be taken to reduce its impact Based on texture analysis, which is crucial for image interpretation in the biomedical domain. We propose a fresh approach for classifying medical images in this context using bidimensional empirical multimodal decomposition (BEMD), this approach is based on the fractal analysis of BIMFs. BEMD is an extension of the one-dimensional case because it has proven to be an adaptive way to represent non-stationary and non-linear signals. Its application to image processing breaks down a image into the total of a number of hierarchical elements “bidimensional intrinsic mode functions (BIMFs)” and residues and the decomposition procedure is iterative. In order to objectively assess the effectiveness of the various BIMF modes and to characterize two states: osteoporotic and healthy, the fractal dimension was calculated for each BIMF using the DBC “Differential Box Counting” method. This novel strategy was applied on a database of CT-Scan medical images of bone textures which contains images of normal and pathological cases. Experimental results indicate that the third mode BIMF achieves higher separation rates compared to the other mode between normal and osteoporotic cases. We use classification rate evaluation criteria, such that the classification rate is given by KNN