Abstract
Currently, there are two kinds of image modalities in clinical diagnosis: the structural and functional images. The x-ray computed tomography (CT) and magnetic resonance image(MRI) are structural images. Single photon emission computed tomography(SPECT) and positron emission tomography (PET),commonly used in Nuclear Medicine, are functional images. Fusion of these two different types of images resolves the problem of insufficient information provided by single modality. Doctor can get more complete information from patients. Registration, using stereotactic frames or skin markers, is complexity and inconvenient. In this study, registration is performed between internal marker in the images. The rotation and translation parameters between the two modalities are solved from neuron network.