Abstracts

Rationale: Surgical management of patients with cingulate epilepsy (CE) is highly challenging, especially in the setting of negative MRI. This study aims to use a voxel-based MRI post-processing technique, implemented in an updated morphometric analysis program (MAP18), to facilitate detection of epileptogenic lesions in CE with a negative pre-surgical MRI.  Methods: Included in this retrospective study were 6 patients with negative MRI and 3 patients with subtly lesional MRI with cingulate epilepsy who underwent surgery and became seizure-free or had marked improvement with 1-year follow-up from the Cleveland Clinic (January 2008 to December 2016) and Second Affiliated Hospital of Zhejiang University (September 2012 to December 2016) surgical databases. Morphometric analysis was applied to pre-surgical T1-weighted volumetric sequence using a fully automated MATLAB script (MAP18), which was then co-registered with other imaging modalities. We analyzed the z-score feature maps from MAP18, as well as the automatic focal cortical dysplasia (FCD) detection probability map by artificial neural networks. Patients were investigated in terms of their image post-processing and functional neuroimaging characteristics, electro-clinical characteristics obtained from noninvasive and invasive evaluations, surgical outcome and pathology. Results: Single MAP+ abnormalities were found in 6 patients (4 patients with anterior cingulate epilepsy; 2 patients with mid cingulate epilepsy), including three MRI-negative and all three patients with subtly lesional MRI. Out of these 6 MAP+ patients, 4 patients became seizure-free after complete resection of the abnormality. Two MAP+ patients did not become seizure free: P1 experienced seizure recurrence at 15 months after laser ablation of the MAP+ abnormality, and subsequently became seizure-free after a second resective surgery which cleaned up the laser ablation margins. P2 achieved marked improvement in seizure frequency and intensity (Class IIb); and the laser ablation area only partially contained the MAP+ abnormality. Automatic detection was successful in two patients; both had subtly lesional MRI. All MAP+ foci were concordant with intracranial EEG when performed. The 3 MAP-negative patients became seizure-free (one Class Ia, two Class Ib) after resection guided by intracranial EEG (each lesion was located in different parts of cingulate gyrus). The localization value of FDG-PET, SPECT and MEG was limited in this cohort. FCD was identified in all patients’ surgical pathology except for P1 and P2 who had laser ablation and therefore surgical specimen was not available. Conclusions: This is the largest study of patients with surgically confirmed MRI-negative cingulate epilepsy (anterior-mid-posterior). MAP was helpful in identifying subtle epileptogenic abnormalities in the majority of the patients included in this study. The new automated FCD detection algorithm showed promising detection results but still needed further improvement. Funding: None