Figure 2
From: DELMEP: a deep learning algorithm for automated annotation of motor evoked potential latencies
Workflow for automated assessment of MEP latencies in a possible closed-loop TMS set-up. MEPs are measured with electrodes placed on the target muscle and stored in a 120-dimension vector. The pre-processing is done by trimming, smoothing, centering, and normalizing the MEP. The resulting vector is used as an input to the neural network for the latency estimation. The dashed arrows show how our DELMEP algorithm could be applied to a closed-loop protocol (dashed box), in which the brain stimulation parameters are modified depending on the MEP responses.
