Pattern Recognition, International Conference on (2006)
Aug. 20, 2006 to Aug. 24, 2006
DOI Bookmark: http://doi.ieeecomputersociety.org/10.1109/ICPR.2006.869
Zhenghui Hu , Hong Kong University of Science and Technology, Kowloon, Hong Kong
Pengcheng Shi , Southern Medical University, Guangzhou, China
For functional magnetic resonance imaging (fMRI) time series data, traditional intensity normalization techniques may introduce negative correlation with the neurological stimulation in non-activated voxels, and hence may cause incorrect identification of the activated/deactivited region. In this study, we present a modified proportional scaling method for intensity normalization using segmented specific tissue. In particular, the mean intensity across the classified cerebrospinal fluid (CSF) cluster, instead of the one across the entire intracerebral voxels, is used for the rescaling of all voxel intensity of a particular image frame. The usefulness of the method is demonstrated on block design fMRI data, which shows that the approach can avoid the negative shift in Z statistics quite well. In addition, this strategy can also be applicable to the analysis of positron emission tomography (PET), single photon emission computed tomography (SPECT) and other functional imaging modalities.
Z. Hu and P. Shi, "Normalization of Functional Magnetic Resonance Images by Classified Cerebrospinal Fluid Cluster," 2006 18th International Conference on Pattern Recognition(ICPR), Hong Kong, 2006, pp. 938-941.