Hsian-Min Chen1, Englin Wong2, Clayton Chi-Chang Chen3, Jyh-Wen Chai3, Shih-Yu Chen4, Ching-Wen Yang5, San-Kan Lee3, Chein-I Chang2
1Department of Biomedical Engineering, HungKuang University, Taichung, Taiwan; 2Department of Computer Science & Electrical Engineering, University of Maryland, Baltimore, United States; 3Department of Radiology, Taichung Veterans General Hospital, Taichung, Taiwan; 4Department of Electrical Engineering, National Chung Hsing University, Taichung, Taiwan; 5Computer Center, Taichung Veterans General Hospital, Taichung, Taiwan
This paper develops an intra-voxel multispectral approach for brain MRI to estimating abundance fractions of tissue substances present in a voxel of an image slice. The idea is originated from remote sensing image processing by using LSMA method to estimate the abundance fraction of a particular tissue substance. The 3D ROC analysis is also developed for performance evaluation where a third dimension is introduced to threshold abundance fractions so as to detection rates. The experimental results demonstrate the utility of the intra-voxel multispectral techniques to compute partial volumes of each of tissue substances via their estimated abundance fractions.