papersSEP 10 04:00 UTC
Deep Learning Method Corrects Temporal Drift in Ultrasound-Based Myocardial Strain Tracking
A new preprint introduces a deep learning approach for tracking heart muscle motion in ultrasound images that embeds physiological constraints into the model. The design is intended to reduce temporal drift across the cardiac cycle, improving the reliability of strain measurements used to assess cardiac function. The work appears on arXiv as paper 2609.09577.