Key result
The proposed CSRNet framework outperformed state-of-the-art methods in quantifying left ventricular metrics, achieving the lowest average Mean Absolute Errors of 134 mm2, 1.54 mm, and 1.16 mm for areas, dimensions, and thicknesses, respectively.
Why the study?
Accurate estimation of full LV metrics from cardiac MR images is challenging due to highly variable cardiac structures across subjects.
Does the CSRNet deep learning framework improve the quantification accuracy of full left ventricular metrics from cardiac MR images compared to state-of-the-art methods?
Does the CSRNet deep learning framework improve the quantification accuracy of full left ventricular metrics from cardiac MR images compared to state-of-the-art methods?
The proposed CSRNet deep learning framework significantly improves the accuracy of quantifying full left ventricular metrics from cardiac MR images compared to existing state-of-the-art methods.
May aid LV quantification accuracy in research settings; leaves open clinical validation before adoption.
Quantifying full left ventricular (LV) metrics including cavity area, myocardium area, cavity dimensions and wall thicknesses from cardiac magnetic resonance (MR) images, and then assessing regional and global cardiac function plays a crucial role in clinical practice. However, due to highly variable cardiac structures across different subjects, it is challenging to obtain an accurate estimation of LV metrics. In this paper, we propose a novel deep learning framework, called cascaded segmentation and regression network (CSRNet), to improve the quantification results. The CSRNet consists of two components: a segmentation component and a regression component. The segmentation component yields myocardial contours of the left ventricle from the input cardiac MR images, and then the regression component learns hierarchical representations from the segmented images and estimates the desired LV metrics. By introducing the myocardial contours, the regression component can pay more attention to the left ventricle, which contributes to more accurate quantification results, although the cardiac structures are variable. The extensive experiments on a dataset of 145 subjects demonstrate that our framework outperforms the state-of-the-art methods.
No takes yet. Share an insight, caveat, or question.
Wang et al. (2019) studied Diverse cardiac pathologies including regional wall motion abnormalities, myocardial hypertrophy, and atrial septal defect (n=145). Cascaded Segmentation and Regression Network (CSRNet) vs. State-of-the-art methods (e.g., DMTRL, FullLVNet, Indices-Net) was evaluated on Mean Absolute Error (MAE) and Correlation Coefficient (ρ) for left ventricular metrics (areas, dimensions, and wall thicknesses). The proposed CSRNet framework outperformed state-of-the-art methods in quantifying left ventricular metrics, achieving the lowest average Mean Absolute Errors of 134 mm2, 1.54 mm, and 1.16 mm for areas, dimensions, and thicknesses, respectively.
Synapse has enriched 5 closely related papers on similar clinical questions. Consider them for comparative context: