Key result
Bayesian-optimized neural networks accurately segment LV and LA echocardiograms with ~0.95 mean Dice overlap.
Why the study?
Bayesian Optimization is an efficient hyperparameter tuning technique, but its application to optimize deep fully convolutional neural networks for multi-structure echocardiography segmentation needed evaluation.
Does Bayesian Optimization improve the segmentation accuracy of a deep fully convolutional neural network for 2D echocardiography compared to a previously published heuristically tuned model?
Does Bayesian Optimization improve the segmentation accuracy of a deep fully convolutional neural network for 2D echocardiography compared to a previously published heuristically tuned model?
Absolute Event Rate: 0.95% vs 0.921%
Bayesian optimization significantly improves the accuracy and clinical reliability of deep learning models for automated 2D echocardiography segmentation.
May support automated echocardiography segmentation; hypothesis-generating and requires prospective validation before clinical use.
Bayesian Optimization (BO) is a well-studied hyperparameter tuning technique that is more efficient than grid search for high-cost, high-parameter machine learning problems. Echocardiography is a ubiquitous modality for evaluating heart structure and function in cardiology. In this work, we use BO to optimize the architectural and training-related hyperparameters of a previously published deep fully convolutional neural network model for multi-structure segmentation in echocardiography. In a fair comparison, the resulting model outperforms this recent state-of-the-art on the annotated CAMUS dataset in both apical two- and four-chamber echo views. We report mean Dice overlaps of 0.95, 0.96, and 0.93 on left ventricular (LV) endocardium, LV epicardium, and left atrium respectively. We also observe significant improvement in derived clinical indices, including smaller median absolute errors for LV end-diastolic volume (4.9mL vs. 6.7), end-systolic volume (3.1mL vs. 5.2), and ejection fraction (2.6% vs. 3.7); and much tighter limits of agreement, which were already within inter-rater variability for non-contrast echo. These results demonstrate the benefits of BO for echocardiography segmentation over a recent state-of-the-art framework, although validation using large-scale independent clinical data is required.
No takes yet. Share an insight, caveat, or question.
Tran et al. (2021) studied Echocardiography segmentation (n=450). Bayesian Optimization of U-net hyperparameters vs. Heuristically found hyperparameters was evaluated on Mean Dice overlap for left ventricular endocardium in apical 2-chamber view. Bayesian Optimization of a deep fully convolutional neural network for echocardiography segmentation achieved mean Dice overlaps of 0.95, 0.96, and 0.93 on LV endocardium, LV epicardium, and left atrium.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: