MELD-graph FCD Detection: Comparing 3T and 7T MRI Sensitivity
- Sensitivity was 65% on 3T scans, 48% on default 7T, and 52% with 7T adaptations.
- The test used 23 patients with FCD and 39 healthy controls from four epilepsy centres.
- Specificity dropped at 7T, and the study is a preprint, not peer‑reviewed.
- MELD‑graph may help flag lesions but still misses many, especially on 7T.
- Further multi‑site trials are needed before clinical adoption.
How does focal cortical dysplasia MRI detection vary by field strength?
The team at King's College London reported that MELD‑graph identified focal cortical dysplasia in 15 of 23 patients on 3T MRI (65% sensitivity). On paired 7T scans the default pipeline found 11 lesions (48%), while an adapted 7T workflow with bias‑field correction and ComBat harmonisation raised detection to 12 cases (52%). Specificity was lower at 7T than at 3T. The work appears as a preprint posted Oct 11, 2026 and has not yet undergone peer review.
What are the limitations of the MELD-graph pipeline at 7T?
Researchers gathered paired T1‑weighted (and where possible FLAIR) images from 23 individuals diagnosed with focal cortical dysplasia and 39 healthy volunteers across four epilepsy centres. Manual lesion masks was the reference standard. The open‑source MELD‑graph pipeline was run on the 3T data, on the 7T data with its default settings, and on the 7T data after adding SPM bias‑field correction, SynthStrip skull‑stripping, and a sub‑millimetric FreeSurfer reconstruction. Some runs also applied ComBat to harmonise scanner‑related differences.
Why is 3T MRI currently superior for epilepsy imaging research?
Because the manuscript is a preprint, the findings have not been vetted by independent reviewers. The sample size is modest, especially for a condition as heterogeneous as focal cortical dysplasia. Sensitivity and specificity are reported, but no information on predictive values or clinical outcomes is provided. The drop in performance at 7T may reflect the current pipeline rather than an inherent limitation of high‑field MRI.
How do MELD-graph MRI findings impact epilepsy patient care?
If MELD‑graph can be tuned for 7T, it might add a second look at scans that appear normal on 3T, potentially catching lesions missed by the eye. However, with detection rates hovering around half of the cases at 7T, relying on the software alone would leave many patients undiagnosed. Clinicians should treat the tool as an adjunct, not a replacement for expert radiological review.
Future directions for MELD-graph epilepsy research
The authors suggest expanding the cohort to include more histologically confirmed cases and testing the pipeline on additional 7T scanners. Prospective studies that track surgical outcomes after MELD‑graph‑guided resections would clarify whether the software improves seizure control. Integration with clinical workflow will also require user‑friendly interfaces and regulatory assessment.
Clinical implications of MELD-graph MRI field strength
At present, MELD‑graph offers modest help on standard 3T MRI but shows reduced reliability on 7T. Until larger, peer‑reviewed studies confirm its performance, clinicians should continue to depend on conventional visual assessment and consider the software as a supplemental opinion.
- Multi-centre validation of MELD-graph for automated focal cortical dysplasia detection using 7T compared to paired 3T MRI — School of Biomedical Engineering & Imaging Sciences, King's College London, UK, Oct 11, 2026
- Multi-centre validation of MELD-graph for automated focal cortical dysplasia detection using 7T compared to paired 3T MRI — medRxiv, Oct 11, 2026
Add NewsPulse Time as a preferred source on Google
Frequently asked questions
The MELD-graph (Morphological Analysis of Epilepsy Lesions) pipeline is an automated AI tool designed to detect focal cortical dysplasia (FCD) in MRI scans, helping clinicians identify potential seizure-causing lesions in patients with epilepsy.
Current research indicates that while 7T MRI offers higher resolution, the MELD-graph pipeline is currently better optimized for the signal-to-noise characteristics and standardized data available at 3T, leading to higher sensitivity in lesion detection.
7T MRI provides superior anatomical detail that can be beneficial for specific clinical cases; however, for automated pipelines like MELD-graph, 3T remains the current standard for reliable, high-sensitivity FCD detection.
By providing objective, automated detection of cortical abnormalities, MELD-graph helps clinicians pinpoint potential surgical targets, reducing the time and subjectivity involved in manual MRI interpretation.
