Analysis of FDG-PET Images for Cancer Diagnosis based on Anatomical Information of Human Body

T. Tozaki, M. Senda, S. Sakamoto, and K. Matsumoto (Japan)


FDG-PET, whole-body voxel model, anatomical informa tion, global registration, local registration


In this paper, we propose the image analysis method of FDG-PET in consideration of anatomical information of human body. As the anatomical information, we use a whole-body voxel model of Japanese adult offered by the public research organization in Japan. The advantage of using human anatomical information is that we can ana lyze the FDG accumulation tendency of each organs with out the dose problem. This method consists of two parts. The one is global registration and the other is local regis tration. In global registration, we search the correspond ing slice image between PET and whole-body voxel model using the characteristic part of human body. As the char acteristics part of human body, we use head, neck, breast and abdomen. In local registration, we apply the follow ing process. The first, we extract the region of interests by thresholding technique. We decide the thresholding value automatically by discriminant analysis method. The sec ond, we apply image fusion technique to these region of interests, and get metamorphosed images. And we show the availability of these method by showing some results.

Important Links:

Go Back