Search

Magnetic resonance imaging has revolutionized the detection of structural abnormalities in patients with epilepsy. However, many focal abnormalities remain undetected in routine visual inspection. Here we use an automated, surface-based method for quantifying morphometric features related to epileptogenic cortical malformations to detect abnormal cortical thickness and blurred gray-white matter boundaries. Using MRI morphometry at 3T with surface-based spherical averaging techniques that precisely align anatomical structures between individual brains, we compared single patients with known lesions to a large normal control group to detect clusters of abnormal cortical thickness, gray-white matter contrast, local gyrification, sulcal depth, jacobian distance and curvature. To assess the effects of threshold and smoothing on detection sensitivity and specificity, we systematically varied these parameters with different thresholds and smoothing levels. To test the effectiveness of the technique to detect lesions of epileptogenic character, we compared the detected structural abnormalities to expert-tracings, intracranial EEG, pathology and surgical outcome in a homogeneous patient sample. With optimal parameters and by combining thickness and GWC, the surface-based detection method identified 92% of cortical lesions (sensitivity) with few false positives (96% specificity), successfully discriminating patients from controls 94% of the time. The detected structural abnormalities were related to the seizure onset zones, abnormal histology and positive outcome in all surgical patients. However, the method failed to adequately describe lesion extent in most cases. Automated surface-based MRI morphometry, if used with optimized parameters, may be a valuable additional clinical tool to improve the detection of subtle or previously occult malformations and therefore could improve identification of patients with intractable focal epilepsy who may benefit from surgery.

The application of advances in biomedical computing to medical imaging research is enabling scientists to conduct quantitative clinical imaging studies using data collected across multiple sites to test new hypotheses on larger cohorts, increasing the power to detect subtle effects. Given that many research groups have valuable existing (legacy) data, one goal of the Morphometry Biomedical Informatics Research Network (BIRN) Testbed is to assess the feasibility of pooled analyses of legacy structural neuroimaging data in normal aging and Alzheimer's disease. The present study examined whether such data could be meaningfully reanalyzed as a larger combined data set by using rigorous data curation, image analysis, and statistical modeling methods; in this case, to test the hypothesis that hippocampal volume decreases with age and to investigate findings of hippocampal asymmetry. This report describes our work with legacy T1-weighted magnetic resonance (MR) and demographic data related to normal aging that have been shared through the BIRN by three research sites. Results suggest that, in the present application, legacy MR data from multiple sites can be pooled to investigate questions of scientific interest. In particular, statistical analyses suggested that a mixed-effects model employing site as a random effect best fits the data, accounting for site-specific effects while taking advantage of expected comparability of age-related effects. In the combined sample from three sites, significant age-related decline of hippocampal volume and right-dominant hippocampal asymmetry were detected in healthy elderly controls. These expected findings support the feasibility of combining legacy data to investigate novel scientific questions.

Due to the increasing need for subject privacy, the ability to deidentify structural MR images so that they do not provide full facial detail is desirable. A program was developed that uses models of nonbrain structures for removing potentially identifying facial features. When a novel image is presented, the optimal linear transform is computed for the input volume (Fischl et al. [2002]: Neuron 33:341-355; Fischl et al. [2004]: Neuroimage 23 (Suppl 1):S69-S84). A brain mask is constructed by forming the union of all voxels with nonzero probability of being brain and then morphologically dilated. All voxels outside the mask with a nonzero probability of being a facial feature are set to 0. The algorithm was applied to 342 datasets that included two different T1-weighted pulse sequences and four different diagnoses (depressed, Alzheimer's, and elderly and young control groups). Visual inspection showed none had brain tissue removed. In a detailed analysis of the impact of defacing on skull-stripping, 16 datasets were bias corrected with N3 (Sled et al. [1998]: IEEE Trans Med Imaging 17:87-97), defaced, and then skull-stripped using either a hybrid watershed algorithm (Ségonne et al. [2004]: Neuroimage 22:1060-1075, in FreeSurfer) or Brain Surface Extractor (Sandor and Leahy [1997]: IEEE Trans Med Imaging 16:41-54; Shattuck et al. [2001]: Neuroimage 13:856-876); defacing did not appreciably influence the outcome of skull-stripping. Results suggested that the automatic defacing algorithm is robust, efficiently removes nonbrain tissue, and does not unduly influence the outcome of the processing methods utilized; in some cases, skull-stripping was improved. Analyses support this algorithm as a viable method to allow data sharing with minimal data alteration within large-scale multisite projects.