Abstract

Focal cortical dysplasia (FCD), a malformation of cortical development, is an important cause of pharmacoresistant epilepsy. Small FCD lesions are difficult to distinguish from normal cortex and remain often overlooked on radiological MRI inspection. This paper presents a method to detect small FCD lesions on T1-MRI relying on surface-based features that model their textural and morphometric characteristics. The automatic detection was performed by a two step classification. First, a vertex-wise classifier based on a neural-network bagging trained on manual labels. Then, a cluster-wise classification designed to remove false positive clusters. The method was tested on 19 patients with small FCD. At the first classification step, 18/19 (95%) lesions were detected. The second classification step kept 13/19 (68%) lesions and decreased efficiently the amount of false positive. This new approach may assist the presurgical evaluation of patients with intractable epilepsy, especially those with unremarkable MRI findings.