Abstract

Quantitative analysis of MR images is becoming increasingly important in clinical trials in multiple sclerosis (MS). This paper describes a fully automated atlas-based technique for segmenting MS lesions from large data sets of multi-channel MR images. The method simultaneously estimates the parameters of a stochastic model for normal brain MR images, and detects MS lesions as voxels that are not well explained by the model. It corrects for MR field inhomogeneities, estimates tissuespecific intensity models from the data itself, and incorporates contextual information in the MS lesion segmentation using a Markov random field. The results of the automated method were compared with lesions delineated by human experts, showing a high total lesion load correlation. When the degree of spatial correspondence between segmentations was taken into account, considerable disagreement was revealed, both between the expert manual segmentations, and between expert and automatic measurements.

Citations