Development of a new statistical evaluation method for Brain SPECT images

Ryuta Kawashima, Kazunori Sato, Hiroshi Ito, Masamichi Koyama, Ryoui Goto, Seiro Yoshioka, Shuichi Ono, Tachio Sato, Hiroshi Fukuda

Research output: Contribution to journalArticlepeer-review


The purpose of this study was to develop a new statistical evaluation method for Brain SPECT images. First, we made normal brain image databases using 99mTc-ECD and SPECT in 10 normal subjects as described previously. Each SPECT images were globally normalized and anatomically standardized to the standard brain shape using Human Brain Atlas (HBA) of Roland et al. and each subject's X-CT. Then, mean and SD images were calculated voxel by voxel. For the next step, 99mTc-ECD SPECT images of a patient were obtained, and global normalization and anatomical standardization were performed as the same way. Then, a statistical map was calculated as following voxel by voxel; (P - Mean)/SD x 10 + 50, where P, Mean and SD indicate voxel value of patient, mean and SD images of normal databases, respectively. We found this statistical map was helpful for clinical diagnosis of brain SPECT studies.

Original languageEnglish
Pages (from-to)1359-1362
Number of pages4
JournalKaku igaku. The Japanese journal of nuclear medicine
Issue number12
Publication statusPublished - 1996


  • Anatomical standardization
  • Cerebral blood flow
  • Human brain
  • Normal database
  • Statistical image


Dive into the research topics of 'Development of a new statistical evaluation method for Brain SPECT images'. Together they form a unique fingerprint.

Cite this