Abstracts

Rationale: Surgical treatment of refractory temporal lobe epilepsy (TLE) presents many challenges in locating the epileptogenic focus and thus not all patients become seizure-free following surgery. Advances in MR imaging techniques can lead to improved localization of the epileptogenic zone and may be validated by correlating MRI with neuropathology of the excised tissue. However, neuropathological features in mildly dysplastic cortical tissue are subtle and difficult to quantify, thus histology image analysis techniques can be applied to characterize these abnormalities. We present a principled computational approach for quantifying neuronal density and neuronal size across cortical laminae and demonstrate its ability to characterize focal regions of mild dysplastic cortex (ILAE Type I/II) in the temporal neocortex. Methods: Patients recommended for anterior temporal lobectomy were recruited and scanned pre-operatively in 3T and 7T MRI. Following surgery, the excised tissue was scanned overnight in 9.4T MRI before histological processing. Histology slides of the excised tissue were cut with a slice thickness of 4 m, immunohistochemically stained with NeuN antibody and scanned at a resolution of 0.5 m. We applied our method to the temporal neocortex from one case (34/M, unilateral MTS, left ATL) in which an experienced neuropathologist labelled regions of normal and abnormal cortex. Neu-N image samples (1mm wide) were drawn perpendicular to the cortical surface and an automated neuron segmentation algorithm (Figure 1) was developed in MATLAB (MathWorks Inc.) to delineate individual neurons. Neuron density and mean neuron size across the cortical lamina were then quantified using overlapping segments from the pial to the white matter boundary. These density and size profiles were evaluated to statistically compare normal and abnormal cortical regions (Figure 2). Results: Normal sections of cortical tissue displayed the expected pattern of lamination with increased density in layers II, IV and VI, and increased size in layers III, V and VI. Abnormal cortical regions had increased density in layer II, decreased density in layers III-VI, and increased neuronal size across all layers. The variability in density was higher than normal regions in layer IV and variability in size was also higher in layers III, V and VI. The increase in size and decrease in density seen in layers III-VI of abnormal cortex highlights one benefit of analyzing individual neuron segmentations, as field fractions or neuron counts alone would be unable to assess that difference.Conclusions: These results suggest that regions of mildly dysplastic cortex can be quantified and compared with our proposed approach, and our ongoing work involves applying this technique to group studies and including additional neuronal features such as orientation and clustering. This work also paves the way for correlative studies with MRI to improve imaging and surgical treatment of cortical dysplasias in refractory epilepsy.