The combined multi-scale deep learning and random forests approach achieved high correlation with manual evaluation for estimating left ventricular volumes, with R values of 0.850 for EDV, 0.871 for ESV, and 0.863 for EF.
Does a combined multi-scale deep learning and random forests approach accurately estimate left ventricular volumes in 3D echocardiography compared to manual evaluation?
A fully learning framework combining multi-scale convolutional deep networks and random forests can directly estimate left ventricular volumes from 3D echocardiography with high correlation to manual evaluation, avoiding the need for segmentation.
Effect estimate: R=0.850 (EDV), 0.871 (ESV), 0.863 (EF)
The experiments results suggested that our proposed method is feasible and can achieve higher accuracy, even in case of echocardiography images with irregular geometry.
Dong et al. (Wed,) conducted a other in Cardiac dysfunction and healthy subjects (n=150). Combined multi-scale convolutional deep network and random forests vs. Ground truth from cardiologist (manual evaluation) was evaluated on Correlation between predicted results and ground truth for EDV, ESV, and EF (R=0.850 (EDV), 0.871 (ESV), 0.863 (EF)). The combined multi-scale deep learning and random forests approach achieved high correlation with manual evaluation for estimating left ventricular volumes, with R values of 0.850 for EDV, 0.871 for ESV, and 0.863 for EF.