Research On Methods Of Multimodality Medical Image Fusion

Multimodality medical image fusion refers to the combination of medical images with different modalities, aiming at fully indicating the advantage the morphological imaging method has of the high resolution and accurate location, overcoming the disadvantage of the low definition in spatial resolution and tissue contrast resolution existing in functional imaging, excavating imaging information to the utmost, and obtaining more information so as to understand the comprehensive information of the diseased tissues or organs, thus providing reliable basis for doctors to make an accurate diagnosis or appropriate treatment program. With a wide application prospect, the medical image fusion has currently become a hotspot in medical image processing. Multimodality medical image fusion is studied in this paper.Aiming at the loss of part of edges and the blur of textures in medical image fusion method which is based on wavelet transform, the method based on multiwavelet transform and fuzzy reasoning is proposed, which uses the multiwavelet with compact support, symmetry and high vanish square to provide a more accurate multi-resolution analysis space for fusion. Upon the analysis of and the comparison between different multiwavelet, the multiwavelet fusion operator most suitable for medical image fusion is selected. In the design of fusion rules, in high-frequency components rules based on fuzzy reasoning are used. The multiwavelet coefficients in high-frequency domain are mapped into fuzzy sets so as to effectively avoid the ambiguity in the fusion process. In low-frequency components, regional weighted variance rules are used. The experiment results show that the proposed method can adequately keep the source image information and has a better fusion performance than wavelets-based fusion method.D-S (Dempster-Shafer) evidence theory has currently been successfully applied to areas of data fusion, risk assessment and surface investigation. In this paper, D-S evidence theory is introduced into the multimodality medical image fusion firstly, and the method for multimodality medical image fusion based on the improved D-S evidence theory is proposed. Firstly, evidence combination rules in D-S evidence theory are improved to deal with the high-evidence conflict; Secondly, with the properties of textures and edges as evidences, the fusion rules are determined according to properties of each point in original images. The experiment results indicate that edges and textures of the original medical image can be relatively completely reserved. The proposed method is better than other methods in fusion performance, and has the characteristic of generality.Finally, the brief summary of this paper is given and the future direction of the research is presented.