Virtual Definition of Tissue by Cluster Analysis of Multi-parametric MR imaging (Virtual-dot-com imaging)
Yovel Y, Assaf Y
Tel Aviv University
Resolution and contrast limits of MRI restrict our current ability to segment and characterize certain tissues. In this study we devised an algorithm (virtual-dot-com imaging) that segments brain tissue based on contrast differences from a multi-parametric MRI protocol. The algorithm uses contrast dynamic range stretching and cluster analysis to segment the tissue to statistically distinct clusters. Using this approach we were able to segment the thalamus to 7 clusters similar to histological atlas. Each cluster could be characterized with a unique contrast finger print. This methodology has the potential for indirect detection of the myelo- and cyto-architecture properties of tissues.