Search

Introduction Seizure semiology is important in the evaluation of patients with drug resistant focal epilepsy to help lateralise and localise the presumed epileptogenic zone (EZ) for curative resection. The localising values of initial semiology are widely variable. Even strongly localising semiology can localise to disparate regions and the symptomatogenic zone can be discordant with the EZ. We created the Semiology-to-Brain Database and 3D Visualisation Tool (SVT) to objectively and directly localise the EZ to potentially aid surgical resection. Methods We curated a database linking seizure semiology to the EZ from an individual-participant systematic review of semiologies, as per PRISMA guidelines. This yielded 11230 localising and 2391 lateralising semiology datapoints from 4643 individual patients across 309 included studies. We included descriptions of the patients’ semiologies and age according to strict ground-truth criteria: if they had a resection and remained seizure free for 12 months, or had stereotactic EEG or stimulation, or concordant neuroimaging and neurophysiology. Localising datapoints were entered into the database under predefined hierarchical lobar structures. We developed a taxonomy of semiological terms, mapped reported brain regions to 3D brain GIF parcellations and lateralising possibilities (contra- or ipsi- lateral, dominant or non-dominant hemisphere). To mitigate publication bias that favours temporal lobe epilepsy, we devised a Bayesian filter to exclude data from studies that preselected patients based on prior topological knowledge of the EZ. We integrated the database, ground truths, mappings, Bayesian filters, and taxonomy into SVT within the freely available 3D-Slicer software platform, enabling visualisations of semiologies. We normalised datapoints so as to set the unit of analysis to a single semiology; such that the sum of all localising datapoints for a single semiology from a single patient adds up to 1. All analyses and code were in python, are open source and freely available on https://github.com/thenineteen/Semiology-Visualisation-Tool. We queried the database to determine the localising probabilities of the 9 most frequently occurring semiological categories, with proportions as point estimates and 95% confidence intervals. We used SVT to make glyph representations of the localising distributions of the entire corpus of semiology. Results Sankey plot of the database with pseudo-glyph representations of bias are shown in Fig. 1. The probabilistic landscape for the 9 most common semiologies are shown in Fig. 2. Conclusions We have curated the largest evidence-based semiology database in the literature and present the probabilistic localising landscape of semiology, which has not been attempted previously. This data-driven approach can be used as clinical decision support for the presurgical evaluation of patients with drug-resistant focal epilepsy, and as the basis for more complex multimodal models for the estimation of the epileptogenic zone.

Poster presented at the American Epilepsy Society Conference 2020. The abstract is pasted below.RationaleIntracranial electroencephalography (icEEG) is used in epilepsy surgery evaluation to define the seizure onset zone. icEEG targets are chosen by clinicians after interpretation of non-invasive data, particularly seizure semiology, scalp EEG and MRI. There are variations in implantation strategies between centres, based on views regarding the relationship between semiological features and brain regions involved.The aim of our research is to create an objective, evidence-based dataset of seizure semiology and the neuroanatomical correlates to inform optimal icEEG implantation strategies.We present an interactive visualisation tool to display the link between clinical semiology and cerebral regions involved.MethodsA literature review of 1208 manuscripts was used to generate a Semiology-to-Brain database called ‘SemioBrain’, containing 13285 data points from 4454 patients. Each data point corresponds to one patient presenting a specific seizure semiology, associated to a certain brain region, based on one or more of the following criteria: postoperative seizure freedom, icEEG monitoring, electrical stimulation, concordance of neuroimaging and neurophysiology.We developed a Python module for the 3D Slicer software platform that references our database to visualise brain regions associated with a set of user-specified semiologies. To search the database, the user can enter custom semiology terms or select from a list of 47 pre-defined terms. Each query can be altered to filter manuscripts according to the ground-truth criteria used to establish the epileptogenic zone and the inclusion criteria for the study, to reduce publication bias.Once the corresponding brain regions are identified, a visualisation is displayed (Fig. 2). An MNI template subdivided into the Neuromorphometrics atlas labels with the geodesic information flows algorithm is used for visualisation.The search result is shown in a table containing each brain region and the corresponding number of patients presenting the pre-selected semiology localising to that region. For visualisation, a score is computed for each region by combining data points across semiologies and is linearly mapped to a colour scheme in which brighter colours and higher opacity correspond to more data points. Brain regions with no involvement in the semiologies are shown either in grey (3D) or not displayed (2D).For more details on the individual structures, the user can place the mouse cursor over the region to visualise the name of the region and corresponding score (Fig. 2). If there is more than one semiological feature, the implicated anatomy can be displayed separately or combined into a composite figure.ResultsCode and documentation are available at https://github.com/thenineteen/Semiology-Visualisation-Tool.ConclusionWe present a data-driven open-source tool to visualise brain regions associated with a set of seizure semiologies. This tool can be used as a clinical decision support system to determine the appropriate strategy for icEEG implantations.

Resective surgery may be curative for drug-resistant focal epilepsy, but only 40% to 70% of patients achieve seizure freedom after surgery. Retrospective quantitative analysis could elucidate patterns in resected structures and patient outcomes to improve resective surgery. However, the resection cavity must first be segmented on the postoperative MR image. Convolutional neural networks (CNNs) are the state-of-the-art image segmentation technique, but require large amounts of annotated data for training. Annotation of medical images is a time-consuming process requiring highly-trained raters, and often suffering from high inter-rater variability. Self-supervised learning can be used to generate training instances from unlabeled data. We developed an algorithm to simulate resections on preoperative MR images. We curated a new dataset, EPISURG, comprising 431 postoperative and 269 preoperative MR images from 431 patients who underwent resective surgery. In addition to EPISURG, we used three public datasets comprising 1813 preoperative MR images for training. We trained a 3D CNN on artificially resected images created on the fly during training, using images from 1) EPISURG, 2) public datasets and 3) both. To evaluate trained models, we calculate Dice score (DSC) between model segmentations and 200 manual annotations performed by three human raters. The model trained on data with manual annotations obtained a median (interquartile range) DSC of 65.3 (30.6). The DSC of our best-performing model, trained with no manual annotations, is 81.7 (14.2). For comparison, inter-rater agreement between human annotators was 84.0 (9.9). We demonstrate a training method for CNNs using simulated resection cavities that can accurately segment real resection cavities, without manual annotations., 13 pages, 6 figures, accepted at the International Conference on Medical Image Computing and Computer Assisted Intervention (MICCAI) 2020

Semiology describes the evolution of symptoms and signs during epileptic seizures and contributes to the evaluation of individuals with focal drug-resistant epilepsy for curative resection. Semiology varies in complexity from elementary sensorimotor seizures arising from primary cortex to complex behaviours and automatisms emerging from distributed cerebral networks. Detailed semiology interpreted by expert epileptologists may point towards the likely site of seizure onset, but this process is subjective. No study has captured the variances in semiological localizing values in a data-driven manner to allow objective and probabilistic determinations of implicated networks and nodes. We curated an open data set from the epilepsy literature, in accordance with Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines, linking semiology to hierarchical brain localizations. A total of 11 230 data points were collected from 4643 patients across 309 articles, labelled using ground truths (postoperative seizure-freedom, concordance of imaging and neurophysiology, and/or invasive EEG) and a designation method that distinguished between semiologies arising from a predefined cortical region and descriptions of neuroanatomical localizations responsible for generating a particular semiology. This allowed us to mitigate temporal lobe publication bias by filtering studies that preselected patients based on prior knowledge of their seizure foci. Using this data set, we describe the probabilistic landscape of semiological localizing values as forest plots at the resolution of seven major brain regions: temporal, frontal, cingulate, parietal, occipital, insula, and hypothalamus, and five temporal subregions. We evaluated the intrinsic value of any one semiology over all other ictal manifestations. For example, epigastric auras implicated the temporal lobe with 83% probability when not accounting for the publication bias that favoured temporal lobe epilepsies. Unbiased results for a prior distribution of cortical localizations revised the prevalence of temporal lobe epilepsies from 66% to 44%. Therefore, knowledge about the presence of epigastric auras updates localization to the temporal lobe with an odds ratio (OR) of 2.4 [CI95% (1.9, 2.9); and specifically, mesial temporal structures OR: 2.8 (2.3, 2.9)], attesting the value of epigastric auras. As a further example, although head version is thought to implicate the frontal lobes, it did not add localizing value compared with the prior distribution of cortical localizations [OR: 0.9 (0.7, 1.2)]. Objectification of the localizing values of the 12 most common semiologies provides a complementary view of brain dysfunction to that of lesion-deficit mappings, as instead of linking brain regions to phenotypic-deficits, semiological phenotypes are linked back to brain sources. This work enables coupling of seizure propagation with ictal manifestations, and clinical support algorithms for localizing seizure phenotypes.