Sensor-free physiological guidance for free-breathing cardiac cine reconstruction using CineJENSE: example of an arrhythmic patient

Motivation: Explore the estimation of physiological signals using the thermal noise variance of radiofrequency coils to reconstruct free-breathing cardiac cine images. Goal(s): Evaluate the reliability of noise variance estimations for ungated free-breathing reconstruction. Approach: Free-breathing reconstructions with reference signals (ECG and respiratory belts) were compared to sensor-free reconstructions using the CineJENSE algorithm with implicit neural representations. The motion estimation and reconstruction for an arrhythmic patient with a bigeminy were examined. Results: The correlation coefficient for the respiration and the F1-score of the cardiac peak detections were computed for quantitative evaluation and showed good concordance with the reference signals, even for the arrhythmic patient. Impact: Sensor-free physiological signal estimations using noise variance from k-space data for free-breathing ungated cardiac cine reconstruction were shown for two examples (healthy volunteer and arrhythmic patient). Results showed good concordance with reference signals and the reconstructions could be performed.