Skip to main content
Fig. 4 | Journal of Cardiovascular Magnetic Resonance

Fig. 4

From: Imaging gravity-induced lung water redistribution with automated inline processing at 0.55 T cardiovascular magnetic resonance

Fig. 4

The automated image processing pipeline. 1, 2 The acquired image was reconstructed inline using the Gadgetron image reconstruction framework. 3 Lung segmentation (red) was performed using a trained U-Net. A region-of-interest (ROI) in the liver was automatically placed under the right lung (white circle) for quantitative purposes. The positioning of the liver ROI was determined from the centroid of the right lung segmentation (blue circle), with the liver ROI was placed in the coronal slice passing through the centroid (blue dashed line). The positioning of the liver ROI was placed 8.75 mm below the bottom of the in-plane lung segmentation (black dashed line) in the centroid column (blue line). 4 Sample of a coil spatially normalized resliced sagittal slice of a healthy subject in the supine position. 5 Quantitative pixel-wise lung water density (LWD) maps relative 70% of the signal intensity in the liver were derived. To enable regional analysis, anterior, mid, and posterior segments were defined (black), that were derived by assigning each segment an equal number of coronal slices. 6 Reconstructed images were streamed back and displayed to CMR host, including a LWD map overlayed on the acquired image

Back to article page