Key result
Fully automated deep learning ventricular segmentation strongly correlated with manual measurements for left ventricular end-systolic volume (r=0.99, P<0.001), but requires expert supervision.
Why the study?
To evaluate the performance of a deep learning algorithm for clinical measurement of right and left ventricular volume and function across cardiac MR images.
Does automated deep learning-based contouring accurately measure right and left ventricular volume and function compared to manual contouring in clinical cardiac MRI examinations?
Population
200 noncongenital clinical cardiac MRI examinations
Comparison
Automated DL-based contouring vs manual contouring
Design
Retrospective validation study
Authors
Loading...
May accelerate supervised CMR volumetry; extends validation data but leaves open unsupervised deployment.
Observational (n=200)
Does automated deep learning-based contouring accurately measure right and left ventricular volume and function compared to manual contouring in clinical cardiac MRI examinations?
Effect estimate: Pearson r = 0.99
p-value: p=< .001
Fully automated ventricular segmentation by deep learning provides accurate contours and volumes but benefits from expert supervision, particularly to resolve errors at the basal and apical slices.
Retson et al. (2020) conducted an observational in Cardiac MRI indications (n=200). Automated deep learning (DL) algorithm for cardiac ventricular volumetry vs. Manual contouring was evaluated on Correlation of left ventricular end-systolic volume (ESV) (Pearson r = 0.99, p=< .001). Fully automated deep learning ventricular segmentation strongly correlated with manual measurements for left ventricular end-systolic volume (r=0.99, P<0.001), but requires expert supervision.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: