Identification of active lesions is critical for the management of multiple sclerosis (MS) patients. Currently this identification is based on post-contrast T1-weighted magnetic resonance imaging (MRI). However, there are safety concerns with repeated administration of gadolinium ?based contrast agents (GBCAs). Thus, there is critical need for identifying active lesions without the use of GBCA. In this application we propose to identify the active lesions without administering GBCA using texture analysis (TA) using multi- modal non-contrast MRI and support vector machine (SVM) learning. A unique feature of this proposal is that the results will be analyzed using MRI data acquired on a large cohort of MS patients (~1000) as a part of phase III, randomized, double-blinded clinical trial (CombiRx). In addition texture features will be identified that can predict lesions that convert into tissue destructive lesions, so called black holes. This has important clinical implications since there is correlative evidence that balk holes are associated with clinical disability. A novelty of this project lies in performing texture analysis in real time that allows the physician to make the decision about administering GBCA on the spot while the patient is still in the scanner. This greatly helps in eliminating and/or minimizing the number of times GBCA needs to be administered. For real time analysis, the necessary infrastructure that includes automatic processing pipeline and integration of the MRI scanner with high performance computational resources located at Texas Advanced Computing Center (TACC) in Austin. Finally to establish real time TA as a viable alternative to GBCA administration for identifying active lesions, the developed methods will be prospectively applied to MS patients undergoing MRI scans as a part of routine clinical management.
Gadolinium based contrast agents (GBCAs) are routinely used for identifying active lesions in multiple sclerosis patients. However, there are safety concerns with repeated administration of GBCA. This proposal uses novel real time analysis for identifying lesions that are likely to enhance on pre- contrast MRI. Considering the large number of MS patients who are frequently scanned with GBCA, this proposal has a high degree of clinical relevance.
