Exploring the discrimination power of the time domain for segmentation andcharacterization of active lesions in serial MR data

This paper presents a new method for the automatic segmentation and charact erization of object changes in time series of three-dimensional data sets. The technique was inspired by procedures developed for analysis of function al MRI data sets. After precise registration of serial volume data sets to 4-D data, we applied a time series analysis taking into account the charact eristic time function of variable lesions. The images were preprocessed wit h a correction of image field inhomogeneities and a normalization of the br ightness over the whole time series, Thus, static regions remain unchanged over time, whereas changes in tissue characteristics produce typical intens ity variations in the voxel's time series. A set of features was derived fr om the times series, expressing probabilities for membership to the sought structures. These multiple sources of uncertain evidence were combined to a single evidence value using Dempster-Shafer's theory. The project was driv en by the objective of improving the segmentation and characterization of w hite matter lesions in serial MR data of multiple sclerosis patients. Pharm aceutical research and patient follow-up requires efficient and robust meth ods with a high degree of automation. The new approach replaces conventiona l segmentation of series of 3-D data sets by a 1-D processing of the tempor al change at each voxel in the 4-D image data set. The new method has been applied to a total of 11 time series from different patient studies, coveri ng time resolutions of 12 and 24 data sets over a period of about 1 year. T he results demonstrate that time evolution is a highly sensitive feature fo r detection of fluctuating structures. (C) 2000 Elsevier Science BN, All ri ghts reserved.