Search

Background and ObjectiveTo test the hypothesis that a multicenter-validated computer deep learning algorithm detects MRI-negative focal cortical dysplasia (FCD).MethodsWe used clinically acquired 3-dimensional (3D) T1-weighted and 3D fluid-attenuated inversion recovery MRI of 148 patients (median age 23 years [range 2–55 years]; 47% female) with histologically verified FCD at 9 centers to train a deep convolutional neural network (CNN) classifier. Images were initially deemed MRI-negative in 51% of patients, in whom intracranial EEG determined the focus. For risk stratification, the CNN incorporated bayesian uncertainty estimation as a measure of confidence. To evaluate performance, detection maps were compared to expert FCD manual labels. Sensitivity was tested in an independent cohort of 23 cases with FCD (13 ± 10 years). Applying the algorithm to 42 healthy controls and 89 controls with temporal lobe epilepsy disease tested specificity.ResultsOverall sensitivity was 93% (137 of 148 FCD detected) using a leave-one-site-out cross-validation, with an average of 6 false positives per patient. Sensitivity in MRI-negative FCD was 85%. In 73% of patients, the FCD was among the clusters with the highest confidence; in half, it ranked the highest. Sensitivity in the independent cohort was 83% (19 of 23; average of 5 false positives per patient). Specificity was 89% in healthy and disease controls.DiscussionThis first multicenter-validated deep learning detection algorithm yields the highest sensitivity to date in MRI-negative FCD. By pairing predictions with risk stratification, this classifier may assist clinicians in adjusting hypotheses relative to other tests, increasing diagnostic confidence. Moreover, generalizability across age and MRI hardware makes this approach ideal for presurgical evaluation of MRI-negative epilepsy.Classification of EvidenceThis study provides Class III evidence that deep learning on multimodal MRI accurately identifies FCD in patients with epilepsy initially diagnosed as MRI negative.

Objective Temporal lobe epilepsy (TLE) is classically associated with hippocampal pathology, and increasingly with structural changes beyond the mesiotemporal lobe. Functional anomalies, their link to cortical morphology and microstructure, and their utility in surgical prognosis, however, remain less well understood. Methods We studied 30 drug-resistant TLE patients and 57 healthy controls using multimodal 3T MRI. All patients had histologically-verified hippocampal sclerosis and underwent postoperative imaging to establish the extent of surgical resection. Our analysis leveraged a novel resting-state fMRI framework that parameterizes functional connectivity distance, consolidating functional and geometric properties of macroscale brain networks. Functional findings were integrated with morphological and microstructural metrics, and associations to post-surgical seizure outcome were examined using machine learning techniques. Results Compared to controls, TLE showed connectivity distance reductions in temporo-insular and dorsomedial prefrontal networks, with the former being more marked ipsilateral to the focus. Findings were driven by contractions in long-distance connections together with an excess in short-range connections, suggesting topological segregation of functional networks. Parallel structural profiling indicated that functional connectivity distance reductions in TLE were independent of atrophy but mediated by microstructural changes in the underlying white matter. Patients with persistent seizures showed anomalies a distinct network signature which predicted seizure outcome with 75±4% accuracy (using 5-fold cross-validation). Conclusions Our findings suggest connectivity distance contractions as a pathoconnectomic substrate of TLE. Functional topological isolation may represent a microstructurally mediated network mechanism that tilts the balance towards epileptogenesis in affected networks.

I.AbstractAging is characterised by accumulation of structural and metabolic changes in the brain. Recent studies suggest transmodal brain networks are especially sensitive to aging, which, we hypothesise, may be due to their apical position in the cortical hierarchy. Studying an open-access healthy cohort (n=102, age range = 30-89 years) with MRI and Aβ PET data, we estimated age-related cortical thinning, hippocampal atrophy and Aβ deposition. In addition to carrying out surface-based morphological and metabolic mapping, we stratified effects along neocortical and hippocampal resting-state functional connectome gradients derived from independent datasets. The cortical gradient depicts an axis of functional differentiation from sensory-motor regions to transmodal regions, whereas the hippocampal gradient recapitulates its long-axis. While age-related thinning and increased Aβ deposition occurred across the entire cortical topography, increased Aβ deposition was especially pronounced towards higher-order transmodal regions. Age-related atrophy was greater towards the posterior end of the hippocampal long-axis. No significant effect of age on Aβ deposition in the hippocampus was observed. Imaging markers correlated with behavioural measures of fluid intelligence and episodic memory in a topography-specific manner. Our results strengthen existing evidence of structural and metabolic change in the aging brain and support the use of connectivity gradients as a compact framework to analyse and conceptualize brain-based biomarkers of aging.