A learned MRI model localizes known dysplasia lesions in FLAIR-negative images
Source check: Source checked. Not independent scientific confirmation.
Finding: A convolutional model trained on lesion-image patterns was evaluated on 18 FLAIR-negative images from 12 patients with histologically confirmed focal cortical dysplasia. The source reports localization in 15 of the 18 images. This is a small retrospective lesion-localization result, not validated population screening or proof of improved surgical outcomes.
Technology: Net-Pos convolutional network; activation-maximization pattern localization
Publication: Peer-reviewed paper · Test context: retrospective or benchmark. Catalog categories, not an evidence grade.
- AI contribution
- A six-layer CNN learns lesion-associated image-block features; activation maximization derives representative patterns used to localize suspected lesions in FLAIR-negative images.
- Test and result
- The overall source dataset has 34 FLAIR images from 19 patients; the FLAIR-negative evaluation contains 18 images from 12 patients. CNN training/test image-block sets are described as coming from unique patients. Localization is compared with lesion regions marked using operative locations.
- Limitation / still unknown
- The abstract calls 15/18recall subject-wise, but its denominator is18 images from 12 patients; it is not presented here as 15 patients detected. Surgical/pathology-selected cases and marked lesion references do not establish accuracy in unselected MRI-negative epilepsy patients. The small dataset, image-block construction and modest segmentation overlap limit generalization; no independent clinical deployment is shown.
Evidence sources and annotation context
Evidence context (descriptions follow the cited sources; test type is a catalog classification):
Evidence sources: https://aapm.onlinelibrary.wiley.com/doi/full/10.1002/acm2.12985
Source checked · editorial details
The source page was checked against this entry. Independent confirmation is not assessed by this label.
Source type: paper. Paper: not linked in this catalog; this does not mean no paper exists.
Review date: 21 Nov 2026. Review scope: Exact primary publisher title/DOI/date; relevant indexed primary cohort, learned-model and reference-validation Methods/Results and limitations. Fresh Crossref publication-status metadata, no linked update; direct full-page access limited..
No follow-up recorded; this is not a claim that no follow-up exists. Follow-up review scope and date are not recorded. No source-supported relationships recorded.
Editorial assessment: Impact 2/5 · Validation 2/5 · Novelty 2/5. Impact and novelty are editorial judgments; validation follows the annotated test type where one exists. Not an evidence grade. Scoring criteria.