Key result
A novel deep reinforcement learning agent outperformed previous reinforcement learning methods and achieved comparable performance to deep learning baselines for left ventricle segmentation.
Why the study?
Relatively few deep reinforcement learning methods have been proposed for image segmentation, and earlier reinforcement learning-based approaches for left ventricle segmentation relied on threshold learning, causing inaccurate results.
Population
ACDC 2017 dataset and Sunnybrook 2009 dataset
Comparison
Novel deep reinforcement learning agent vs previous reinforcement learning methods and deep learning baselines
Authors
Loading...
May aid LV analysis in data-limited settings; leaves open clinical validation before workflow adoption.
A novel deep reinforcement learning approach for left ventricle segmentation achieves high accuracy even with limited training data, outperforming previous RL methods.
Xiong et al. (2021) studied Left ventricle segmentation. Deep reinforcement learning agent (First-P-Net and Next-P-Net) vs. Previous reinforcement learning methods and deep learning baselines was evaluated on Left ventricle endocardium segmentation performance. A novel deep reinforcement learning agent outperformed previous reinforcement learning methods and achieved comparable performance to deep learning baselines for left ventricle segmentation.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: