ECG-based model predicts liver motion without additional imaging

An ECG-only deep-learning model predicted respiratory-induced liver motion with a mean absolute error of 2.83 mm during normal breathing and 4.02 mm overall, with correlation coefficients above 0.90 in eight volunteers.

• An encoder-decoder network mapped electrocardiogram (ECG) signals directly to internal liver motion, eliminating the need for simultaneous imaging or external respiratory bellows.
• More than 90% of predictions fell within the accepted error margin for needle-insertion procedures, suggesting feasibility for interventional guidance.
• The source acknowledges accuracy declined during deep breathing when liver excursion increased, indicating a current limitation for high-amplitude motion.

Automated summary

RadPigeon summaries are original and for information only. They are not clinical advice.