ملخص: Among the gestures that the doctors achieve regularly, there is an important task of comparing medical images: compare images of the same modality and of the same patient, but acquired at different moments, compare images of the same patient but from different modalities (RMI, Scanner, ...), compare images of different patients, or compare an image with an anatomical or a functional atlas. These tasks of comparison raise all of the same problem: the registration. The term registration means the mapping of the 2D or 3D images provided by the various sensors of imagery, in order to superpose the pixels or voxels representing the same structures. The interest of the registration of the medical images is for example to follow the evolution of a lesion, to evaluate a morphological variation or pathological data, to synthesize additional information resulting from several examinations of the same patient on the same image, to simulate an action, a treatment or to assist a surgery by a computer... A classification of the used medical image registration methods is the one that divides them in two great families: geometrical methods and iconic methods. Another classification is based on criteria formulated by Van Den Elsen, pol. and Viergever. In this work, we are interested to the problem of compensation of the movement of the patient during the acquisition spaced in time of two medical images of the same modality. This kind of problem can be studied like a geometrical problem. We thus used a semi-automatic geometrical method by using for the first time the orthogonal polynomials on 2D images. The results are presented in this paper.

لغة: إنجليزية