Perfusion-based segmentation of the human brain using similarity mapping

In this work, a method for segmenting human brain MR scans on the basis of perfusion is described. This technique uses a measure of similarity between the time-intensity curves obtained with dynamic susceptibility contrast-enhanced MRI and a modeled curve of reference to isolate a tissue of interest, such as white or gray matter. The aim of this study was to validate the method by performing segmentation of white and gray matter in six controls. The relative regional blood volume gray-to-white matter ratio was used as a criterion to assess the quality of segmentation. On average, this ratio was 2.1 ± 0.2, which is in good agreement with the literature, thus suggesting reliable segmentation. In the case of abnormal perfusion, time-intensity curves are different in shape than that of normal tissue. Therefore, this approach might allow the segmentation of pathological regions, and combined with an indicator-dilution analysis might offer new possibilities for characterizing a brain pathology. Magn Reson Med 45:261–268, 2001. © 2001 Wiley-Liss, Inc.