Automated Detection and Surgical Planning for Focal Cortical Dysplasia with Multicenter Validation

Deep Learning Approaches for Imaging-Based Automated Segmentation of Tuberous Sclerosis Complex

The present study presents a novel approach for identifying epileptogenic tubers in patients with tuberous sclerosis complex (TSC) and automating tuber segmentation using a three-dimensional convolutional neural network (3D CNN). The study retrospectively included 31 TSC patients whose lesions were manually annotated from multiparametric neuroimaging data. Epileptogenic tubers were determined via presurgical evaluation and stereoelectroencephalography recording. Neuroimaging metrics were extracted and compared between epileptogenic and non-epileptogenic tubers. Additionally, five datasets with different preprocessing strategies were used to construct and train 3D CNNs for automated tuber segmentation. The normalized positron emission tomography (PET) metabolic value was significantly lower in epileptogenic tubers defined via presurgical evaluation (p = 0.001). The CNNs showed high performance for localizing tubers, with an accuracy between 0.992 and 0.994 across the five datasets. The automated segmentations were highly correlated with clinician-based features. The neuroimaging characteristics for epileptogenic tubers were demonstrated, increasing surgical confidence in clinical practice. The validated deep learning detection algorithm yielded a high performance in determining tubers with an excellent agreement with reference clinician-based segmentation. Collectively, when coupled with our investigation of minimal input requirements, the approach outlined in this study represents a clinically invaluable tool for the management of TSC.

Magnetic resonance-guided laser interstitial thermal therapy vs. open surgery for drug-resistant mesial temporal lobe epilepsy: a propensity score matched retrospective cohort study

BACKGROUND

Magnetic resonance-guided laser interstitial thermal therapy (MRgLITT) and traditional open surgery (OS) are effective and safe options for patients with drug-resistant mesial temporal lobe epilepsy (DR-mTLE). However, their superiority in seizure control and preservation of functional abilities remains unclear. This study aimed to compare the surgical outcomes of MRgLITT and OS.

MATERIALS AND METHODS

This multicenter retrospective cohort study included patients with DR-mTLE who underwent MRgLITT or OS at three centres between 2015 and 2023. The data on patient demographics, presurgical non-invasive evaluation, stereoelectroencephalography (SEEG) implantation, memory alteration, and seizure outcomes were collected. Propensity score matching (PSM) analysis was conducted for the comparison of seizure control and functional preservation between two surgical approaches.

RESULTS

Of the 244 individuals who met the study criteria, 33 underwent MRgLITT and 211 OS. The median (interquartile range) age at seizure onset was 22.0 (13.0) and 12.3 (10.0) years in the MRgLITT and OS groups, respectively. The first PSM, based on demographic and non-invasive information, resulted in 26 matched pairs for the primary analysis. There were no significant differences in memory preservation ( P = 0.95) or surgical outcomes ( P = 0.96) between the groups. The second PSM, based on demographics and SEEG implantation, yielded 32 matched pairs for the sensitivity analysis, showing similar results. Subset analysis of early and late MRgLITT cases revealed no statistically significant differences in the proportion of patients with memory decline ( P = 0.42) or seizure control ( P = 1.00). Patients who underwent SEEG implantation were 96% less likely to achieve seizure freedom after MRgLITT ( P = 0.02).

CONCLUSION

Minimally invasive MRgLITT is associated with memory preservation and seizure control, similar to traditional OS. MRgLITT is effective and safe for DR-mTLE and is relevant for future prospective randomized trials on dominant-side mTLE, providing practical implications for guiding neurosurgeons in the selection of surgical approaches.

Multimodal imaging-based diagnostic approach for MRI-negative posterior cortex epilepsy

Background

Posterior cortex epilepsy (PCE) primarily comprises seizures originating from the occipital, parietal, and/or posterior edge of the temporal lobe. Electroclinical dissociation and subtle imaging representation render the diagnosis of PCE challenging. Improved methods for accurately identifying patients with PCE are necessary.

Objectives

To develop a novel voxel-based image postprocessing method for better visual identification of the neuroimaging abnormalities associated with PCE.

Design

Multicenter, retrospective study.

Methods

Clinical and imaging features of 165 patients with PCE were retrospectively reviewed and collected from five epilepsy centers. A total of 37 patients (32.4% female, 20.2 ± 8.9 years old) with magnetic resonance imaging (MRI)-negative PCE were finally included for analysis. Image postprocessing features were calculated over a neighborhood for each voxel in the multimodality data. The postprocessed maps comprised structural deformation, hyperintense signal, and hypometabolism. Five raters from three different centers were blinded to the clinical diagnosis and determined the neuroimaging abnormalities in the postprocessed maps.

Results

The average accuracy of correct identification was 55.7% (range from 43.2 to 62.2%) and correct lateralization was 74.1% (range from 64.9 to 81.1%). The Cronbach's alpha was 0.766 for the correct identification and 0.683 for the correct lateralization with similar results of the interclass correlation coefficient, thus indicating reliable agreement between the raters.

Conclusion

The image postprocessing method developed in this study can potentially improve the visual detection of MRI-negative PCE. The technique could lead to an increase in the number of patients with PCE who could benefit from the surgery.