Takeuchi Ryo, Sengoku Tami, Matsumura Kaname
Shizuoka PET Imaging Center, 4-27-1 Kita-anndoh, Aoi-ku, Shizuoka 420-8527, Japan.
Radiat Med. 2006 Aug;24(7):538-44. doi: 10.1007/s11604-006-0054-x.
The positioning or selection of a region of interest (ROI) is an essential step for the quantification of brain images. However, so long as the ROIs were manually selected, results obtained fluctuated considerably with subtle changes in their positioning. To perform an ROI analysis of the brain with improved objectivity and excellent reproducibility, we recently established fully automated ROI-based analysis software for the brain, the so-called 3DSRT and the FineSRT. 3DSRT performed by consecutive analyzing process as follows: (1) anatomical standardization using the SPM99 algorithm; (2) analysis using constant 318 ROIs divided into 12 groups (segments) on each hemisphere; (3) calculation of the area-weighted average for each of the respective 24 segments; (4) display of the results followed by the saving of respective values of the 636 ROIs (both hemispheres). The processes of the FineSRT were fundamentally similar to that of 3DSRT, but on FineSRT more precise constant 1394 ROIs (both hemispheres) corresponding to the respective cerebral convolutions were used. Our programs have been useful for objective estimation of cerebrovascular reserve, especially for follow-up studies of an infant with moyamoya disease, because we could compare the results without consideration of the patient's growth during the follow-up period.
