Lost in Tracking: Uncertainty-guided Cardiac Cine MRI Segmentation at Right Ventricle Base

Accurate biventricular segmentation of cardiac magnetic resonance (CMR) cine images is essential for the clinical evaluation of heart function. Deep-learning-based methods have achieved highly accurate segmentation performance, however, compared to left ventricle (LV), right ventricle (RV) segmentation is still more challenging and less reproducible. The degenerated performance frequently occurs at the RV base, where the in-plane anatomical structures are complex (with atria, valve, and aorta), and varying due to the strong inter-planar motion. In this work, we propose to tackle the currently unsolved issues in CMR segmentation, specifically at the RV base, with two strategies: first, we complemented the public resource by re-annotating the RV base in the ACDC dataset, with refined delineation of the right ventricle outflow tract (RVOT), under the guidance of an expert cardiologist. Second, we proposed a novel Dual-Encoder U-Net architecture that leverages temporal incoherence to inform the segmentation when inter-planar motions occur. The inter-planar motion is characterized by loss-of-tracking, via Bayesian uncertainty of a motion-tracking model. Our experiments showed that our method significantly improved the RV base segmentation by taking temporal incoherence into account. Additionally, we investigated the reproducibility of deep-learning-based segmentation and showed that the combination of consistent annotation and loss-of-tracking could enhance RV segmentation reproducibility, potentially facilitating a large number of clinical studies focusing on RV.