Influence of cerebral perfusion template on statistical parametric mapping in SPECT imaging

• Main Authors: Bang-Hung Yang, 楊邦宏
• Other Authors: Jyh-Cheng Chen
• Format: Others
• Language: zh-TW
• Published: 2006
• Online Access: http://ndltd.ncl.edu.tw/handle/79183711881076686491

碩士 === 國立陽明大學 === 放射醫學科學研究所 === 94 === Objective In vivo stability of tracer distribution and the availability of 99mTc-ECD SPECT images for healthy individuals are critical to the successful application of regional cerebral blood flow (rCBF) in clinical neurological studies. Statistical parametric mapping (SPM) software has been designed for voxel-by-voxel analysis of the whole brain imaging. Spatial normalization is the essential preprocessing step to transform individual images into the MNI standard template which corresponds with Talairach anatomic coordinates. However, the SPECT standard template provided with SPM2 software is 99mTc-HMPAO brain perfusion image. The purpose of this study was to evaluate the outcome of templates on healthy brain perfusion analysis, and to compare four kinds of spatial normalization template to see whether it will influence statistical results. In addition, we tried to vary the size of smoothing filter, which was used to increase signal to noise ratio, in order to make choice of optimal size of smoothing filter for imaging analysis. Material and methods There were fifteen healthy volunteers recruited. Performances of two 99mTc-ECD SPECT scans on the same subject about 2 weeks apart are required for assessment of brain perfusion changes over time. MRI T1 images were done between two examinations of the 99mTc-ECD SPECT images. We utilized four kinds of spatial normalization template in the outcome of statistical analysis of 99mTc-ECD SPECT studies: (1) the standard SPM2 SPECT (99mTc- HMPAO) template; and (2) an averaged 99mTc-ECD SPECT template; and (3) MRI anatomic template normalization approach ; and (4) an MRI-based 99mTc-ECD SPECT template. Results and conclusions After applying four kinds of spatial normalization template, SPM results show that the MRI anatomic template normalization approach is more reliable than the standard SPM2 SPECT template for neurological study. In addition, the optimal size of smoothing kernel for improving SNR was demonstrated within the range of 3 times to 4 times of pixel size in SPECT analysis.