One outstanding challenge for machine learning in diagnostic biomedical imaging is algorithm interpretability. A key application is the identification of subtle epileptogenic focal cortical dysplasias (FCDs) from structural MRI. FCDs are difficult to visualize on structural MRI but are often amenable to surgical resection. We aimed to develop an open-source, interpretable, surface-based machine-learning algorithm to automatically identify FCDs on heterogeneous structural MRI data from epilepsy surgery centres worldwide. The Multi-centre Epilepsy Lesion Detection (MELD) Project collated and harmonized a retrospective MRI cohort of 1015 participants, 618 patients with focal FCD-related epilepsy and 397 controls, from 22 epilepsy centres worldwide. We created a neural network for FCD detection based on 33 surface-based features. The network was trained and cross-validated on 50% of the total cohort and tested on the remaining 50% as well as on 2 independent test sites. Multidimensional feature analysis and integrated gradient saliencies were used to interrogate network performance. Our pipeline outputs individual patient reports, which identify the location of predicted lesions, alongside their imaging features and relative saliency to the classifier. On a restricted ‘gold-standard’ subcohort of seizure-free patients with FCD type IIB who had T1 and fluid-attenuated inversion recovery MRI data, the MELD FCD surface-based algorithm had a sensitivity of 85%. Across the entire withheld test cohort the sensitivity was 59% and specificity was 54%. After including a border zone around lesions, to account for uncertainty around the borders of manually delineated lesion masks, the sensitivity was 67%. This multicentre, multinational study with open access protocols and code has developed a robust and interpretable machine-learning algorithm for automated detection of focal cortical dysplasias, giving physicians greater confidence in the identification of subtle MRI lesions in individuals with epilepsy.

INTRODUCTION: Since its first description in the early 19 th century, the inferior frontooccipital fascicle (IFOF) and its anatomo-functional features were neglected in the neuroscientific literature for the last century. In the last decade, the rapid development of in vivo imaging for the reconstruction of white matter (WM) connectivity (i.e., tractography) and the consequent interest in more traditional ex vivo methods (postmortem dissection) have allowed a renewed debate about course, termination territories, anatomical relationships, and functional roles of this fascicle.EVIDENCE ACQUISITION: We reviewed the main current knowledge concerning the structural and functional anatomy of the IFOF and possible implications in neurosurgical practice.EVIDENCE SYNTHESIS: The IFOF connects the occipital cortex, the temporo-basal areas, the superior parietal lobule, and the pre-cuneus to the frontal lobe, passing through the ventral third of subinsular WM of the external capsule. This wide distribution of cortical terminations provides multimodal integration between several functional networks, including language, non-verbal semantic processing, object identification, visuo-spatial processing and planning, reading, facial expression recognition, memory and conceptualization, emotional and neuropsychological behavior. This anatomo-functional organization has important implication also in neurosurgical practice, especially when approaching the frontal, insular, temporo-parieto-occipital regions and the ventricular system.CONCLUSIONS: The IFOF is the most extensive associative bundle of the human connectome. Its multi-layer organization reflects important implications in many aspects of brain functional processing. Accurate awareness of IFOF functional anatomy and integration between multimodal datasets coming from different sources has crucial implications for both neuroscientific knowledge and quality of neurosurgical treatments.

OBJECTIVE Lateral periinsular hemispherotomy (LPH) and vertical parasagittal hemispherotomy (VPH) are the most popular disconnective techniques for intractable epilepsies associated with unilateral hemispheric pathologies. The authors aimed to investigate possible differences in seizure outcome and complication rates between patients who underwent LPH and VPH. METHODS A comprehensive literature search of PubMed and Embase identified English-language articles published from database inception to December 2019 that reported series (minimum 12 patients with follow-up ≥ 12 months) on either LPH or VPH. Pooled rates of seizure freedom and complications (with a particular focus on hydrocephalus) were analyzed using meta-analysis to calculate both fixed and random effects. Heterogeneity (Cochran’s Q test) and inconsistency (fraction of Q due to actual heterogeneity) were also calculated. RESULTS Twenty-five studies were included. Data from 825 patients were available for seizure outcome analysis (583 underwent LPH and 242 underwent VPH), and data from 692 patients were available for complication analysis (453 underwent LPH and 239 underwent VPH). No differences were found in the pooled rates of Engel class I seizure outcome between patients who underwent LPH (80.02% and 79.44% with fixed and random effects, respectively) and VPH (79.89% and 80.69% with fixed and random effects, respectively) (p = 0.953). No differences were observed in the pooled rates of shunted hydrocephalus between patients who underwent LPH (11.34% and 10.63% with fixed and random effects, respectively) and VPH (11.07% and 9.98% with fixed and random effects, respectively) (p = 0.898). Significant heterogeneity and moderate inconsistency were determined for hydrocephalus occurrence in patients who underwent both LPH and VPH. CONCLUSIONS LPH and VPH techniques present similar excellent seizure outcomes, with comparable and acceptable safety profiles.