MULTIMODALITY IMAGE REGISTRATION USING TSALLIS ENTROPY IN STATISTICAL PARAMETRIC MAPPING.

Abstract

This project proposes the study of the performance and robustness of the TSallis entropy in multimodality registration images based in the mutual information (MI) to the software Statistical Parametric Mapping (SPM). The objective is to find an alternative co-registration method of different modalities of medical images, once the most of algorithms implemented for such function uses the mutual information theory through the Shannon entropy.The overlapping of different modalities of medical images, technique known as co-registration, can assume an excellent role in the localization of the epileptogenic zone during presurgical workup of patients with pharmacoresistant epilepsy. The present study used images of MRI (Magnetic Resonance Imaging) and ictal and interictal SPECT (Single Photon Emission Computed Tomography), both examples of structural and functional images respectively. Preliminary results indicate that the qualitative analysis of co-registration, and the computer time through the use of the entropy proposal are very similar to the obtained results for the classic entropy. Some tests had disclosed greater robustness of the TSallis entropy in relation to Shannon. Through the comparison of the transformation matrices it will be possible to quantitatively determine the efficiency of the method. For the evaluation of the TSallis entropy in co-registration based in the mutual information theory, we will perform a comparative analysis among multimodality images supplied by the Center for Epilepsy Surgery of the HCFMRP- USP.