Schematic overview of the platform. (IMAGE)
Caption
A sketch of the whole methodology for the continuous monitoring of the organoid synaptic activity through the analysis of neural activity from hFOs using MEA and a GDN model. Neural signals are recorded from organoids cultured in 24-well plates in treated and untreated conditions (e.g., U1M and U1M + VPA) using MEA reading systems. The recorded voltage signals are split into training and test datasets. The GDN training (orange path) involves 3 steps: (1) sensor embeddings are computed to represent the activity of individual electrodes; (2) a graph structure is learned to model dynamic inter-electrode relationships; and (3) a forecasting model predicts future neural signals. Once the GDN is trained, the temporal graph descriptors of the training set are used to build the classification model then validated using the descriptors obtained by test datasets (green path).
Credit
Arianna Mencattini, University of Rome Tor Vergata.
Usage Restrictions
News organizations may use or redistribute this image, with proper attribution, as part of news coverage of this paper only.
License
Licensed content