Meeting Banner
Abstract #1108

Correlations Between Ex-Vivo MRI and Hippocampal Sub-Field Neuronal Density in Temporal Lobe Epilepsy

Maged Goubran1, 2, Terry M. Peters1, 3, Ali R. Khan4

1Robarts Research Institute , London, Ontario, Canada; 2Biomedical Engineering, University of Western Ontario, London , Ontario, Canada; 3Biomedical Engineering, University of Western Ontario, London, Ontario, Canada; 4Robarts Research Institute, London, Ontario, Canada

The identification of cellular abnormalities in histology of the resected tissue has the potential to validate the effectiveness of multi-spectral magnetic resonance imaging (MRI) in pre-operative foci localization in temporal lobe epilepsy. We investigated the correlation between ex-vivo MR images and neuronal cell density of hippocampal sub-fields from histological slides. Segmented cell counts of hippocampal sub-fields were consistent with pathological findings as well as local entropy values of MR images. We have shown that ex-vivo MR images have the potential to differentiate between the hippocampal sub-fields through local features and entropy measurements.

Abstract PDF Poster PDF

Keywords

acceptable addition affected affects agar aligned assessed automated available better biomedical brain cell cellular classification classifier clinically close commonly components computed connected consistent coronal correlation count counts datasets densities density despot detection differentiate drug early effectiveness embedding engineering entropy epilepsy error fast feature features field fields fiesta findings fixed focus focused forest formalin grossing histology identification improve in vivo incorporated institute intensity interactive international intervention intractable invasiveness investigate iterative labeled labels learning lobe local localization loss manually many mapping maps matrix measure measures medically neocortex neurology neuron neuronal normalized operative outcomes pathology patients pattern paving peters pixel planning plots potential preliminary previous probability processed prone quantitative random recruited registered registration resection resistant resolution rigid rigidly risk robust sampled sclerosis sectioned segmentation seizure silicone slice slides solenoid sparsely specimens spectral subfield subfields subject suggested surgical target temporal terry textural texture third thresholding tissue toolkit trees underwent unit validate validation views warped warping western whether