You are viewing the SNL 2017 Archive Website. For the latest information, see the Current Website.

Poster A54, Wednesday, November 8, 10:30 – 11:45 am, Harborview and Loch Raven Ballrooms

Gliosis+ for continuous lesion quantification in VLSM to map brain-language relationships

Lisa Krishnamurthy1,2, Venkatagiri Krishnamurthy2,3, Amy Rodriguez2, Michelle Benjamin4,5, Keith McGregor2,3,5, Atchar Sudhyadhom5, Kaundinya Gopinath6, Bruce Crosson2,3,5,7;1Dept. of Physics & Astronomy, Georgia State University, Atlanta, GA, United States, 2Center for Visual and Neurocognitive Rehabilitation, Atlanta VAMC, Decatur, GA, United States, 3Dept. of Neurology, Emory University, Atlanta, GA, United States, 4Brooks Rehabilitation, Jacksonville, FL, United States, 5University of Florida, Gainesville, FL, United States, 6Dept. of Radiology & Imaging Sciences, Emory University, Atlanta, GA, United States, 7Dept. of Psychology, Georgia State University, Atlanta, GA, United States

Combining structural imaging with language behavior is emerging as a powerful clinical research tool to better understand stroke-related language deficits (aphasia). One method used for this purpose, Voxel-based Lesion Symptom Mapping (VLSM), is a simple yet elegant method used to define structure-behavior associations, but depends on binary “all-or-nothing” lesion masks. In 2005, Tyler et al. advanced VLSM by discarding the binary lesion masking step by means of correlating the continuous T1-weighted image signal intensity with continuous language measures, and demonstrated that T1-image signal intensity has high correlations with word processing abilities. Unfortunately, the T1-image normalization process is directly influenced by subject-specific lesion size, atrophy, and ventricle size, such that this normalization adds noise into the analysis. To address these issues, we have developed a VLSM-regression technique that relates a novel continuous MRI signal quantity (gliosis+) with continuous measures of behavior, but is free from MRI signal processing artifacts. Fourteen monolingual English speaking subjects were recruited >6 months post left-hemisphere stroke with evidence of non-fluent language output. The Western Aphasia Battery (WAB) and Boston Naming Test (BNT) were administered to assess verbal fluency, repetition, comprehension, and word retrieval in all subjects. High-resolution T1-weighted MPRAGE and T2SPACE structural images were acquired on each subject. The ratio of T1-MPRAGE to T2SPACE was calculated, and inverted to increase sensitivity to the lesion and surrounding structural changes. Images were spatially normalized to MNI space, and the anterior portion of the lateral ventricle and grey matter were segmented in the non-lesioned hemisphere to calculate upper and lower bounds of the gliosis+ maps. The T2/T1 ratio was then subdivided into ten compartments based on the upper and lower bounds, and automatically segmented, depicting the smooth and continuous transition from cavitation to surrounding glial and axonal damage. VLSM-regression with the behavioral score was calculated for all voxels that had a gliosis+ score for at least 4 or more subjects to ensure a good fit. Maps were thresholded at p=0.05, and clustered at 60 voxels to obtain regions of significant relationship with clinical measures of language behavior. All 14 subjects had quantifiable gliosis+ scores within the left hemisphere, and gliosis+ score location corresponded to the location of the lesion and surrounding structural damage. Linear regression revealed significant relationships between gliosis+ and WAB fluency, WAB repetition, WAB comprehension, and BNT. As an example, the gliosis+ score was related to WAB Repetition in a significant cluster of voxels in supramarginal gyrus and posterior arcuate fasciculus (R2=0.52). To the best of our knowledge, this work presents the first evidence that gliosis+ maps provide a measure that can be related to clinical measures of language behavior in patients with aphasia. The VLSM-regression results were sensitive to expected language areas of importance, and were able to account for a large variability in language behavior measures. These preliminary data show that we will be able to readily extend the gliosis+ VLSM-regression analysis to a larger cohort to obtain a robust comparison of gliosis+ VLSM-regression to standard VLSM.

Topic Area: Methods

Back to Poster Schedule