Intelligent in-cell electrophysiology: Reconstructing intracellular action potentials using a physics-informed deep...

Area of Science:

• Electrophysiology
• Cardiology
• Neuroscience
• Pharmacology

Background:

• Traditional electrophysiology methods like patch-clamp are precise but limit throughput and are invasive.
• Nanoelectrode Arrays (NEAs) offer high-throughput simultaneous intracellular and extracellular recordings but face challenges in accessing intracellular potentials.
• Developing non-invasive, high-throughput methods is crucial for advancing cellular electrical property studies.

Purpose of the Study:

• To develop an AI-supported technique for reconstructing intracellular action potentials (iAPs) from extracellular recordings.
• To establish extracellular signals as reliable indicators of intracellular activity.
• To enable non-invasive, high-throughput drug cardiotoxicity assessments.

Main Methods:

• Utilized thousands of synchronous extracellular action potential (eAP) and iAP pairs from stem-cell-derived cardiomyocytes on NEAs.
• Analyzed correlations between specific eAP and iAP features.
• Developed a physics-informed deep learning model to reconstruct iAPs from extracellular recordings.

Main Results:

• Demonstrated strong correlations between extracellular and intracellular action potential features.
• Showcased the capability of extracellular signals to reliably indicate intracellular activity.
• Validated the AI model's effectiveness in reconstructing iAP waveforms from NEA and Microelectrode array (MEA) data.

Conclusions:

• AI-driven reconstruction of iAPs from extracellular recordings is feasible and reliable.
• This non-invasive approach facilitates high-throughput drug cardiotoxicity screening.
• The developed model holds significant potential for future electrophysiology research across diverse cell types and drug interactions.