Data fusion processing in brain medical imaging

Abstract

During recent years, medical imaging examinations more and more often use information acquired from multiple imaging modalities. This is mainly because the complementary information after fusion of the data, improves the quality of the diagnosis. For instance, the image fusion can be between data from different modalities or different individuals. It may also concern fusion of image data with an external model, which expresses prior knowledge about the problem at hand. Thus to have more information, medical image fusion is done which combines these complimentary features into one image. Generally it involves two steps, registration and fusion Registration deals with proper geometrical alignment of the images so that the corresponding pixels or regions of both images map to the same region being imaged. Registration is necessary in order to be able to compare or integrate the data obtained from different measurements. Therefore, correspondences between the two images are determined. Then, one image can be transformed into the coordinate system of the other. Perfect image registration is required to fuse the images. Image Fusion essentially deals with the integration of information from different images to obtain a single image containing the complete information. In order to generate an overlapping image from different-modality images, image registration and various fusion techniques have been employed. The objective of our work is to implement algorithms for fusing data on Medical images. For that, we have firstly developed registration methods which are: Rigid, Affine, Projective and Elastic. They are used to find a mapping or a transformation of points from one image to the corresponding points in another image. Secondly, we develop and implement Medical Fusion methods which use respectively logical operators, pseudocolormap, and clustering algorithms (k-means, fuzzy k-means, thresholding and EMsegmentation) for extracting and utilising information from several scans simultaneously in the analysis and diagnosis of patients. Finally, we applied the developed methods for different applications such as: Tumor's image extraction to monitor growth of tumor during a period of time, Extraction of normal tissue images and to add them to different modality to get one complete image