Multi-centre validation of MELD-graph for automated focal cortical dysplasia detection using 7T compared to paired 3T MRI
- 1School of Biomedical Engineering & Imaging Sciences, King's College London, UK
Abstract
BackgroundSoftware tools can support MRI diagnosis of focal cortical dysplasias (FCDs) - subtle, surgically-treatable lesions that underlie drug-resistant focal epilepsy (DRFE). MELD-graph, a widely used open-source tool combining surface-based morphometry with a convolutional neural network, was developed using 1.5T and 3T MRI data. 7T MRI is increasingly used for lesion detection in DRFE. This study investigated whether MELD-graph is robust to domain shift at 7T.
MethodsWe collated paired 3T and 7T T1-weighted MRI, and FLAIR where available, from 23 patients with FCD (6 histologically-verified, 17 identified radiologically) and 39 healthy controls at four epilepsy centres. Manual lesion labels were used as ground truth. MELD-graph was run on 3T images, on 7T images with the default pipeline, and on 7T images with adaptations for 7T MRI (SPM bias-field correction, SynthStrip skull-stripping, sub-millimetric FreeSurfer reconstruction), with and without ComBat harmonisation. FCD detection performance and surface morphometry features were compared.
ResultsSensitivity for FCD detection was 65% (15/23) at 3T, 48% (11/23) for 7T default and 52% (12/23) for 7T adapted, with Combat harmonisation. Among radiologically 3T-negative patients, 2/8 were detected at 3T and 3/8 at 7T; and among histologically confirmed FCD, 3/6 at 3T and 1/6 at 7T. Specificity was significantly lower at 7T (default 39%, adapted 45%) than 3T (76%). False-positives predominantly occurred in the basal temporal lobes. ComBat harmonisation improved specificity in all conditions and largely compensated for field-strength dependent differences in surface morphometry features. Adding FLAIR at 7T increased false-positive predictions.
ConclusionSensitivity is comparable but specificity of MELD-graph is lower at 7T compared to 3T. Adaptations for 7T MRI can reduce segmentation errors but do not significantly improve FCD detection, while harmonisation improves performance. The adapted pipeline is released open-source. 7T MELD-graph outputs may inform radiologist review in challenging cases but require cautious interpretation.
Related
- Automated detection of focal cortical dysplasia in epilepsy using machine learning on structural MRI
- Validation of a Revised V-Slope Algorithm for Automated First Ventilatory Threshold Detection in Healthy Adults
- Simulating the Radiological Appearance of Focal Cortical Dysplasia in MRI
- Balloon cells in type II focal cortical dysplasia: developmental origins, molecular signatures, and epileptogenic mechanisms.
- Epileptic hub in MRI-negative focal cortical dysplasia using advanced dynamic statistical parametric mapping and magnetoencephalography.
BibTeX
@misc{kronlage2026multicentre,
title = {{Multi-centre validation of MELD-graph for automated focal cortical dysplasia detection using 7T compared to paired 3T MRI}},
author = {C. Kronlage and M. Ripart and R. J. Piper and C. Casella and K. Klodowski and T. Bauer and M. H. Eriksson and P. Bridgen and M. M. Tisdall and S. R. Bouschery and A. Lange and E. D. Pracht and T. Stocker and J. O. Cleary and F. D'Arco and T. Baldeweg and J. S. Duncan and R. Surges and T. Ruber and C. T. Rodgers and T. E. Cope and D. W. Carmichael and J. O'Muircheartaigh and S. Adler and K. Wagstyl},
howpublished = {medRxiv},
year = {2026},
month = oct,
doi = {10.64898/2026.10.08.26365027},
url = {https://www.medrxiv.org/content/10.64898/2026.10.08.26365027v1},
note = {CC BY},
}
RIS
TY - UNPB
TI - Multi-centre validation of MELD-graph for automated focal cortical dysplasia detection using 7T compared to paired 3T MRI
AU - C. Kronlage
AU - M. Ripart
AU - R. J. Piper
AU - C. Casella
AU - K. Klodowski
AU - T. Bauer
AU - M. H. Eriksson
AU - P. Bridgen
AU - M. M. Tisdall
AU - S. R. Bouschery
AU - A. Lange
AU - E. D. Pracht
AU - T. Stocker
AU - J. O. Cleary
AU - F. D'Arco
AU - T. Baldeweg
AU - J. S. Duncan
AU - R. Surges
AU - T. Ruber
AU - C. T. Rodgers
AU - T. E. Cope
AU - D. W. Carmichael
AU - J. O'Muircheartaigh
AU - S. Adler
AU - K. Wagstyl
PY - 2026
DA - 2026/10/08
PB - medRxiv
DO - 10.64898/2026.10.08.26365027
UR - https://www.medrxiv.org/content/10.64898/2026.10.08.26365027v1
AB - BackgroundSoftware tools can support MRI diagnosis of focal cortical dysplasias (FCDs) - subtle, surgically-treatable lesions that underlie drug-resistant focal epilepsy (DRFE). MELD-graph, a widely used open-source tool combining surface-based morphometry with a convolutional neural network, was developed using 1.5T and 3T MRI data. 7T MRI is increasingly used for lesion detection in DRFE. This study investigated whether MELD-graph is robust to domain shift at 7T. MethodsWe collated paired 3T and 7T T1-weighted MRI, and FLAIR where available, from 23 patients with FCD (6 histologically-verified, 17 identified radiologically) and 39 healthy controls at four epilepsy centres. Manual lesion labels were used as ground truth. MELD-graph was run on 3T images, on 7T images with the default pipeline, and on 7T images with adaptations for 7T MRI (SPM bias-field correction, SynthStrip skull-stripping, sub-millimetric FreeSurfer reconstruction), with and without ComBat harmonisation. FCD detection performance and surface morphometry features were compared. ResultsSensitivity for FCD detection was 65% (15/23) at 3T, 48% (11/23) for 7T default and 52% (12/23) for 7T adapted, with Combat harmonisation. Among radiologically 3T-negative patients, 2/8 were detected at 3T and 3/8 at 7T; and among histologically confirmed FCD, 3/6 at 3T and 1/6 at 7T. Specificity was significantly lower at 7T (default 39%, adapted 45%) than 3T (76%). False-positives predominantly occurred in the basal temporal lobes. ComBat harmonisation improved specificity in all conditions and largely compensated for field-strength dependent differences in surface morphometry features. Adding FLAIR at 7T increased false-positive predictions. ConclusionSensitivity is comparable but specificity of MELD-graph is lower at 7T compared to 3T. Adaptations for 7T MRI can reduce segmentation errors but do not significantly improve FCD detection, while harmonisation improves performance. The adapted pipeline is released open-source. 7T MELD-graph outputs may inform radiologist review in challenging cases but require cautious interpretation.
ER -