Photo credit: www.sciencedaily.com
Researchers have unveiled an innovative AI tool capable of identifying 64% of brain abnormalities related to epilepsy that often elude human radiologists.
The tool, known as MELD Graph, has the potential to transform the healthcare landscape for approximately 30,000 epilepsy patients in the UK and around 4 million globally who suffer from focal cortical dysplasia (FCD), a significant contributor to epilepsy.
A study published in JAMA Neurology by a collaborative team from King’s College London and University College London (UCL) highlights the enhanced capability of this tool in diagnosing FCD, which is among the main causes of epilepsy.
According to the researchers, MELD Graph not only promises to accelerate the diagnostic process but also allows for timely surgical intervention, which could result in cost savings for the NHS of up to £55,000 per patient.
In the UK, the prevalence of epilepsy stands at about 1 in 100 individuals, with one-fifth of those experiencing seizures due to identifiable structural abnormalities in the brain. FCDs are frequently responsible for these seizures, and patients with this condition often find their symptoms unresponsive to medication. Surgical removal of the lesion offers a viable solution to mitigate seizures. However, due to their often subtle appearance, these lesions can frequently escape detection by radiologists, with studies suggesting that up to half of such instances go unnoticed. Consequently, delayed diagnosis and treatment can lead to increased seizure frequency, more emergency department visits, and significant disruption to daily life.
The research involved compiling MRI data from 1,185 participants, including 703 individuals diagnosed with FCD and 482 controls, sourced from 23 epilepsy centers globally as part of the Multicentre Epilepsy Lesion Detection project (MELD). Among the participants, half were children. The AI tool was then trained on these scan results to detect subtle brain abnormalities that might otherwise remain hidden.
Dr. Konrad Wagstyl from King’s College London, the lead author of the study, noted, “Radiologists currently face an overwhelming volume of images to analyze. Implementing an AI tool like MELD Graph can enhance their diagnostic decisions, bolster NHS efficiency, expedite patient treatment timelines, and reduce the necessity for extraneous tests.”
Co-author Dr. Luca Palma from Bambino Gesù Children’s Hospital in Italy shared a case where MELD Graph identified a lesion overlooked by multiple radiologists in a 12-year-old boy enduring daily seizures and who had exhausted nine anti-seizure medications with no positive results. He emphasized that this technology could significantly aid in identifying suitable candidates for surgical interventions, improving surgical planning, minimizing risks, and ultimately enhancing patient outcomes.
Although MELD Graph is not yet in clinical use, the research team has made the AI tool available as open-source software. They are actively conducting workshops to train healthcare professionals globally, including those at Great Ormond Street Hospital and the Cleveland Clinic, on its application.
Dr. Mathilde Ripart from UCL, the first author, expressed pride in the global reach of their tool, noting its impact in assisting doctors across various countries, including the UK, Chile, India, and France.
Co-author Professor Helen Cross, who holds the Prince of Wales’s Chair of Childhood Epilepsy and also serves as a consultant at Great Ormond Street Hospital, highlighted the prolonged challenges many children face in obtaining a proper diagnosis. She pointed out that the epilepsy community is eager to find faster routes to treatment, with initiatives like MELD offering significant promise in swiftly identifying treatable abnormalities.
Dr. Sophie Adler, co-lead from UCL, remarked on the importance of international cooperation in facilitating such research. She acknowledged the contributions of 75 researchers and clinicians working collectively toward the shared vision of eliminating missed epilepsy lesions around the world.
Source
www.sciencedaily.com