The beating heart: artificial intelligence for cardiovascular application in the clinic

Figures

• 

  Fig. 1. Identified fields that artificial intelligence has the potential to impact in cardiovascular magnetic resonance imaging

• 

  Fig. 2. Automatic detection of trigger delay and acquisition window for whole-heart coronary magnetic resonance angiography imaging. A and B represent two different phase suggestions. The position of the coronal imaging slab (in yellow) and isotropic image-navigator box (in blue) are also determined (adapted with permission from Wood et al. [12])

• 

  Fig. 3. Automatic inversion time selection for LGE imaging. The algorithm detects epicardium contours, from which the centroid of the segmentation is extracted. Subsequently, boxes of empirically selected size (orange squares in figure) are created around the centroid and the signal intensity is extracted from the pixels in it

• 

  Fig. 4. Main cardiac magnetic resonance imaging applications where segmentation is needed. Red: endocardium, green: epicardium. Yellow contour: left atrium

• 

  Fig. 5. Fully automated analysis of cardiac T1 mapping (adapted from Fahmy et al. [53])

• 

  Fig. 6. Example of different steps needed in parameter extraction for cardiac magnetic resonance imaging. From left to right: cardiac images are acquired, followed by automated segmentation of various anatomical heart structures, enabling subsequent parameter extraction