Quantitative Automated Lesion Detection of Traumatic Brain Injury

Because of their non-focal nature, TBI-related brain lesions are difficult to detect and quantify with traditional MRI. In the current research program the investigators propose to develop quantitative automated lesion detection (QALD) procedures to (1) clarify the nature and distribution of tissue damage following mild, moderate and severe TBI (2) improve the capability of detecting, quantifying, and localizing TBI brain damage in individual patients and (3) correlate quantitative measures of brain damage in individual TBI patients with neuropsychological deficits in attention, memory, and executive function. QALD detects abnormal tissue parameters in the diseased brain through statistical comparisons with a normative database. Preliminary results show that QALD is capable of detecting highly significant abnormalities in the brains of TBI patients with normal clinical MRI scans. QALD will be further enhanced and tested with a larger database and including brain images acquired with four different imaging sequences (T1, T2, DTI and fluid-attenuated inversion recovery or FLAIR) from 100 control subjects. Data analysis will incorporate advanced cortical surface mapping techniques to quantify gray matter tissue parameters and thickness in 34 distinct cortical regions in each hemisphere. In addition, cortical fiber projections will be quantified with DTI and FLAIR analysis of white matter lying below the cortical surface. Subcortical fiber tracts critical for complex cognitive operations will be analyzed with voxel-based morphometry and with improved region of interest algorithms to define fiber tract boundaries. Tissue properties in critical subcortical structures (e.g., the hippocampus) will be quantified after automatic parcellation of these brain regions. The investigators will also test the control subjects on a battery of neuropsychological tests (NPTs) and correlate variations in the size, myelination, and tissue properties of normal cortical and subcortical structures with cognitive performance. Then, the investigators will gather identical imaging data in 99 TBI patients divided into three groups (mild, moderate and severe TBI) in order to characterize the average pattern of damage caused by TBIs of different severity. Next, the investigators will quantify lesions in individual TBI patients and describe the variability of lesion patterns in the different severity groups. In parallel, the investigators will develop further multimodal analysis techniques to combine statistical information from different imaging sequences to improve lesion-detection sensitivity to co-localized abnormalities evident with different imaging protocols. In addition, the investigators will test patients with NPTs and analyze the relationship between brain damage, cognitive performance and self-assessments of outcome in order to improve the prognostic value of neuroradiological studies of TBI.