The Use of Artificial Intelligence In Magnetic Resonance Imaging of Epilepsy: A Systematic Review and Meta-Analysis
Bibliographic record
Abstract
ABSTRACT Background The application of artificial intelligence (AI)/machine learning (ML) to MRI can be a powerful tool to streamline clinical decision-making, yet variability amongst MRI sequences and algorithms have hindered appropriate assessment of reliability and generalizability. Methods We conducted a systematic review and meta-analysis of the ability of current AI/ML models operating on MRI data for: 1) epilepsy diagnosis, 2) temporal lobe epilepsy lateralization, 3) lesion localization, and 4) post-surgical outcome prediction. Searches were conducted across PubMed, Medline, and Embase databases from inception until January 1, 2025. We selected studies that employed AI/ML models trained on any MRI modality to classify at least ten patients across the four main objectives for qualitative assessment, and further included in the meta-analysis if they reported an accuracy rate. The primary outcome of the meta-analysis was the overall accuracy of AI/ML models trained on MRI data. The secondary outcome was the concomitant risk of bias evaluation using PROBAST. Findings We identified 158 studies for qualitative evaluation and 127 studies for inclusion in the meta-analysis. AI/ML on multimodal MRI could accurately distinguish epilepsy patients from healthy controls (overall accuracy: 88% [85-90]), lateralize temporal lobe epilepsy (90% [87-93]), localize epileptogenic lesions (82% [74-88]), and predict post-surgical seizure-freedom (83% [78-87]). Overall, a high risk of bias remains in the literature; participant bias remained high across all outcomes (64-87%), as well as predictor (88-100%) and analysis (83-100%) bias. Outcome bias was low only for AI/ML studies predicting post-surgical outcomes. Interpretation Our results support promising accuracy of AI/ML models in epilepsy diagnostics and prognostics but remain highly susceptible to bias in participants, predictors, outcome, and analysis domains which limits current translation to routine clinical practice. We encourage closer interdisciplinary collaboration between clinical and scientific groups to improve validation studies based on thorough study design, analysis, and reporting.
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.032 | 0.076 |
| Meta-epidemiology (narrow) | 0.002 | 0.001 |
| Meta-epidemiology (broad) | 0.015 | 0.030 |
| Bibliometrics | 0.008 | 0.009 |
| Science and technology studies | 0.001 | 0.001 |
| Scholarly communication | 0.004 | 0.002 |
| Open science | 0.002 | 0.001 |
| Research integrity | 0.002 | 0.002 |
| Insufficient payload (model declined to judge) | 0.002 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".