Our latest research, published in the Epilepsy Research journal, marks a significant advancement in the detection of Focal Cortical Dysplasias (FCDs), which are localized abnormalities in the cerebral cortex linked to drug-resistant epilepsy. These malformations can interrupt the brain’s normal electrical activity, leading to epilepsy in those affected.

This breakthrough was achieved through a collaborative effort with the Hamburg Epilepsy Center and Epileptologicum Hamburg, resulting in the development of a promising convolutional neural network (CNN) model. This model showcases high sensitivity and specificity in detecting and segmenting FCDs, a crucial factor in the diagnosis and treatment of epilepsy.

One of the standout features of our model is its use of a continuous learning approach. This approach enables the model to continuously improve its performance while processing new clinical MRI data, thereby providing an effective solution to the problem of catastrophic forgetting, which is the phenomenon where a neural network loses previously learned information as soon as it learns new information.

By offering a more accurate method for identifying FCDs, our model paves the way for more targeted and effective treatment options, potentially revolutionizing the approach to epilepsy management.
This publication highlights the promising role of AI in medical diagnostics and opens new avenues for addressing complex neurological conditions.