Biomedical Image Fusion Using Wavelet Transforms and SOFM Neural Network

Use of Computer Tomography(CT) , Magnetic Resonance (MR), Single Photon Emission Computed Tomography (SPECT) techniques in biomedical imaging has revolutionized the process of medical diagnosis in recent years. Further advancements in biomedical imaging are being done by development of new tools of image processing. One of tools is image fusion .The fusion of CT scan , MR and SPECT images can make medical diagnosis much easier and accurate. On the basis of review of research papers published earlier in the area of image fusion, this paper presents a method of image fusion based on Discrete Wavelet Transform(DWT) and Self Organizing Feature Mapping(SOFM) neural network. The proposed method is a feature level fusion method.2-dimensional DWT is used to decompose the images into various details at different levels to extract useful features and SOFM neural network is used to recognize complementary features. These features are then integrated using a criteria based on activity level. Final fused image is constructed from fused feature set. The proposed method of fusion has been applied to MR, SPECT and CT scan images, and results obtained are presented in this paper.