Non-invasive visualization of the living human brain with MRI (magnetic resonance imaging) is an essential part of clinical neurology. Until now, evaluation of structural changes in diseases have mainly relied upon qualitative visual judgements, or tedious hand labeling. This project aims to develop MRI image processing software that will automatically segment the entire brain volume into its constituent tissues and structures. This segmentation allows quantification of tissue volumes and intrinsic parameters. The software will be interoperable with our suite of structural and functional MRI software tools, allowing, for example, subcortical regions of interest to be examined for functional activation. We propose to base the tissue segmentation of intrinsic properties of brain tissue, thus rendering it independent of pulse sequence or scanner manufacturer. This will greatly increase scan/rescan reliability which is crucial for multi-site clinical trials, or for within-subject repeated measures (e.g., comparing tumors or multiple sclerosis plaques over time). The proposed software will allow MRI data to be used to detect neuropsychiatric disease monitor the course of such diseases, and evaluate the effectiveness of treatments. Research and clinical applications can be expected in several neurodegenerative and psychiatric diseases such as schizophrenia, Alzheimer's and Huntington's disease.
The proposed product would be of interest to researchers and clinicians using MRI, i.e., >10000 scanners worldwide. The proposed software could become a standard part of many brain MRI scans, in order to detect and quantify pathology, and thus could potentially be used in a substantial proportion of the millions of brain scans performed every year. Initially, the software would be used to perform surrogate markers of CNS effectiveness in drug evaluation.
