Novel AI algorithm may be the key for a breakthrough epilepsy treatment
- Scientists from the University College London developed an artificial intelligence (AI) algorithm to detect subtle anomalies in the brain.
- The project gathered more than 1,000 patients’ MRI scans from 22 international epilepsy centers to develop the algorithm.
- The development is specifically important for detecting abnormalities.
A group of scientists from the University College London has developed an artificial intelligence (AI) algorithm that can detect drug-resistant focal cortical dysplasia (FCD), a subtle anomaly in the brain that leads to epileptic seizures. This is a promising step for scientists toward detecting and curing epilepsy in its early stages.
To develop the algorithm, the Multicentre Epilepsy Lesion Detection project (MELD) gathered more than 1,000 patients’ MRI scans from 22 international epilepsy centers, which reports where anomalies are in cases of drug-resistant focal cortical dysplasia (FCD), a major reason behind epilepsy.
FCDs are areas of the brain that have evolved abnormally and therefore cause drug-resistant epilepsy. Generally, its treatment requires surgery. But it's challenging to spot these lesions via MRI since FCDs appear normal in MRI scans.
The team of researchers quantified cortical features from the MRI scans and employed about 300,000 locations across the brain to develop the algorithm. The next step was to train the algorithm on cases categorized by expert radiologists as having a healthy brain or having FCD. The results demonstrated that, on average, the algorithm was successful in identifying the FCD in 67 percent of cases in the cohort (538 participants).
Previous research couldn't uncover the abnormalities in 178 of the participants, which means that radiologists had been unable to find the abnormality. The MELD algorithm, on the other hand, managed to spot the FCD in 63% of these cases.
Paving the way for the cure
This development is important, especially for detecting abnormality.
“We put an emphasis on creating an AI algorithm that was interpretable and could help doctors make decisions. Showing doctors how the MELD algorithm made its predictions was an essential part of that process,” said Mathilde Ripart, the co-first author of the study. "
"This algorithm could help to find more of these hidden lesions in children and adults with epilepsy, and enable more patients with epilepsy to be considered for brain surgery that could cure the epilepsy and improve their cognitive development," added a co-senior author of the study, Dr. Konrad Wagstyl. "Roughly 440 children per year could benefit from epilepsy surgery in England," he added.
The results of the study have been published in the journal Brain.
One outstanding challenge for machine learning in diagnostic biomedical imaging is algorithm interpretability. A key application is the identification of subtle epileptogenic focal cortical dysplasias (FCDs) from structural MRI. FCDs are difficult to visualize on structural MRI but are often amenable to surgical resection. We aimed to develop an open-source, interpretable, surface-based machine-learning algorithm to automatically identify FCDs on heterogeneous structural MRI data from epilepsy surgery centres worldwide.
The Multi-centre Epilepsy Lesion Detection (MELD) Project collated and harmonized a retrospective MRI cohort of 1015 participants, 618 patients with focal FCD-related epilepsy and 397 controls, from 22 epilepsy centres worldwide. We created a neural network for FCD detection based on 33 surface-based features. The network was trained and cross-validated on 50% of the total cohort and tested on the remaining 50% as well as on 2 independent test sites. Multidimensional feature analysis and integrated gradient saliencies were used to interrogate network performance.
Our pipeline outputs individual patient reports, which identify the location of predicted lesions, alongside their imaging features and relative saliency to the classifier. On a restricted ‘gold-standard’ subcohort of seizure-free patients with FCD type IIB who had T1 and fluid-attenuated inversion recovery MRI data, the MELD FCD surface-based algorithm had a sensitivity of 85%. Across the entire withheld test cohort the sensitivity was 59% and specificity was 54%. After including a border zone around lesions, to account for uncertainty around the borders of manually delineated lesion masks, the sensitivity was 67%.
This multicentre, multinational study with open access protocols and code has developed a robust and interpretable machine-learning algorithm for automated detection of focal cortical dysplasias, giving physicians greater confidence in the identification of subtle MRI lesions in individuals with epilepsy.