The Multi-centre Epilepsy Lesion Detection (MELD) project is an international collaboration dedicated to improving the detection of lesions in patients with drug-resistant epilepsy.

We hope you find this site a useful source of information about the MELD project.

Automated Detection of Focal Cortical Dysplasias

Focal cortical dysplasia (FCD) is a congenital abnormality of cortical development and a leading cause of surgically remediable drug resistant epilepsy. MRI has played a major role in the evaluation of patients; yet, significant proportions of lesions remain undetected by conventional image analysis. Machine learning offers a powerful framework to develop automated and individualised clinical tools that may improve the detection of lesions and prediction of clinically relevant outcome.

In work published in Neuroimage: Clinical in 2017, Adler, Wagstyl et al., developed a classifier using surface-based features to identify focal abnormalities of cortical development in a paediatric cohort. Focal cortical dysplasias in this paediatric cohort were correctly identified in 73% of the children.

FCD examples
Examples of cortical area detected by the neural network classifier in 5 patients with a radiological diagnosis of FCD. First column: T1-weighted images. Second column: FLAIR images. White circle on T1 and FLAIR images indicates lesion location. Third column: Neural network classifier output (yellow) and manual lesion mask (light blue) viewed on pial surface, for large lesions, or inflated surface, for small lesions buried in sulci.

Since this publication, a number of groups world-wide have downloaded the freely available code so that they can replicate this work at their own site, on their own scanner and patients. However, to date, surface-based automated lesion detection studies have been single-centre and it is unclear how generalisable the developed frameworks and tools are. Furthermore, machine learning continues to improve with increasing numbers of examples.

The MELD Project is an international collaboration aiming to develop lesion detection and normalisation techniques for the incorporation of data and sequences from multiple sites. One of the major focuses of the project is to create open-access, robust and generalisable tools for FCD detection.

We aim to:

  • continue to develop AI capabilities to detect FCDs
  • train classifiers on data from multiple centres (including paediatric and adult data)
  • create clinically useful tools that can be used in the pre-surgical evaluation of patients with drug-resistant epilepsy to aid radiological diagnosis

Thanks for taking a look and feel free to contact us if you would like to find out more!


You can contact us at MELD.study@gmail.com.

You’re very welcome to contact us via GitHub too.

This website is hosted via GitHub pages. If you see any typos or other mistakes please let us know…or file a pull request with your edits.


This website is based on Kirstie Whitaker’s website. It uses the So Simple Theme by designer slash illustrator Michael Rose. THANK YOU Kirstie and Michael :tada: :sparkles: