Understanding the potential for deep learning algorithms in epilepsy diagnosis

Takeaway

  • Convolutional neural networks (CNN) visualization and deep learning methodologies can successfully aid the diagnosis of histomorphologically difficult-to-treat malformations of cortical development.

Why this matters

    Definitive diagnosis of epileptogenic brain lesions such as focal cortical dysplasia (FCD) type IIb and tuberous sclerosis complex (TSC) based on histopathological findings alone is difficult. This study could help to introduce deep learning applications and visualization to aid the histopathologic diagnosis of rare and difficult-to-classify brain lesions.