A Segmentation Method to Identify MRI and FDG-PET Negative Focal Cortical Dysplasia
Abstract number :
2.220
Submission category :
5. Neuro Imaging
Year :
2015
Submission ID :
2327502
Source :
www.aesnet.org
Presentation date :
12/6/2015 12:00:00 AM
Published date :
Nov 13, 2015, 12:43 PM
Rationale: Focal cortical dysplasias (FCDs) are the most common cause of medically refractory seizures. • They are often small (< 3 cm3). • On MRI they are frequently not visible or are very subtle. • They favor depth-of-gyrus / bottom-of-sulcus localization. • On FDG-PET they appear identical to adjacent white matter. It is possible to expose these lesions by isolating the gray matter portion of coregistered MRI-PET.Methods: As part of an IRB-approved study of postprocessing techniques to identify potential seizure onset zones for epilepsy surgery, FCDs were identified in 12 cases by imaging or pathology. All were initially MRI negative with up to 3 prior MRIs, and were initially PET negative. Imaging postprocessing steps: T1 MRI skull stripped PET coregistration to skull stripped T1 T1 gray matter segmentation Invert PET intensity Remove PET white matter with T1 segmentation mask Coregistered PET gray matter can be reviewed with any other coregistered imaging study.Results: 9 cases had depth of sulcus/gyrus dysplasias in which the abnormality on PET had the imaging appearance of white matter with conventional radiological viewing. Coregistration was essential to identify the hypometabolic cortex. In some cases unsegmented PET overlaid on MRI successfully exposed the abnormality, but required very fine adjustment of transparency, windowing, and color maps. Visual isolation of the segmented gray matter PET volumes more readily identified the abnormal cortex, particularly when used with appropriate color maps. Since windowing of lower intensity features will eliminate rather than emphasize hypometabolism, inverting PET intensities was essential. 3 of these 9 cases had much larger or remote hypometabolic regions that failed to identify a specific area as a dysplasia. With segmentation the more striking depth of sulcus abnormality became apparent. 2 cases had orbital frontal and one had a temporal pole dysplasia not in a depth of gyrus, but which were apparent after PET segmentation. 3 of the 12 cases had dysplasias identified by pathology but not by MRI. 6 cases have proceeded to resective surgery.Conclusions: Bottom of sulcus/ depth of gyrus dysplasias typically appear visually as normally hypometabolic white matter. Segmentation of coregistered PET gray matter, especially if intensity-inverted and viewed with appropriate color maps, can expose cortical dysplasias that otherwise may not appear with conventional radiological viewing. Presurgical identification of dysplasias can improve intracranial EEG localization and can reduce the extent of surgery required for seizure freedom. We are currently developing and evaluating an automated method for performing PET segmentation.
Neuroimaging