Evaluation of a machine-based learning algorithm for detection of focal cortical dysplasia

Takeaway

  • A deep convolutional neural network (CNN) to detect focal cortical dysplasia (FCD) produced both high sensitivity and high specificity findings and was also able to detect magnetic resonance imaging (MRI)-negative FCD with 85% sensitivity.

Why this matters

  • MRI-negative FCD contributes to diagnostic challenges in this disease. Identification of a computer-based deep learning method can increase detection of FCD, leading to earlier intervention and improved outcomes.