Performance Evaluation Method of Image Registration in Diffusion-Tensor Imaging Based on Gray Level Co-Occurrence Matrix

Release time:2021-12-25  Hits:

• Journal:MEDICAL SCIENCE AND BIOLOGICAL ENGINEERING (MSBE2015)
• Key Words:Diffusion-tensor imaging, image registration, evaluation
• Abstract:Medical image registration is the key technique of multimodal medical image processing, because it directly influences the results of information fusion of multimodal signals. It is one of the most important technique in Diffusion-Tensor Imaging (DTI), since the tensor calculation depends on the registration of Diffusion-Weighted Imaging (DWI) of different gradient direction. Therefore, the evaluation of the registration performance is a very important step for evaluating the result of information fusion. But traditional subjective evaluation method varies from person to person, while some objective evaluation methods are related to different registration algorithm or software. To solve this problem, an automatic evaluation method of DWI registration is proposed in this paper. The evaluation criteria used in this method include entropy, contrast, and correlation calculated by gray level co-occurrence matrix of Fractional Anisotropy (FA) image. Two registration results obtained by rigid and affine methods, and non-registered data are used in our testing experiment. The experiment results of 5 subjects data showed that the proposed evaluation method not only can integrate whole-brain information automatically, but also can reflect the performance of different registration methods accurately.
• Co-author:Min Lu, Xiaojie Zhang, Changshui Zhang,Yifei Yang, Weibei Dou
• First Author:Yang Chen,Yanglei Song
• Indexed by:会议论文
• Correspondence Author:Weibei Dou,Changshui Zhang
• Page Number:282-288
• Translation or Not:no
• Date of Publication:2015-05-30