Deep convolutional neural networks for detection of cortical dysplasia: a multicenter validation

摘要: PURPOSEFocal cortical dysplasia (FCD) is a prevalent surgically-amenable epileptogenic malformation of cortical development. On MRI, FCD typically presents with cortical thickening, hyperintensity, and blurring of the gray-white matter interface. These changes may be visible to the naked eye, or subtle and be easily overlooked 1. Despite advances in MRI analytics, current surface-based algorithms 2-5 do not detect FCD in up to 50% of cases 6. We propose a novel algorithm to distinguish FCD from healthy tissue directly on MRI voxels. Our method harnesses feature learning capability of convolutional neural networks (CNN) 7 with minimal data pre-processing. Our algorithm was trained and tested on data from the Montreal Neurological Institute (S1), and tested on independent data from S1 and four sites worldwide (S2-S5), for a total of 185 individuals.