System and method for CNN segmentation for automated key phase detection in dynamic contrast-enhanced liver MRI

A system for selecting reference images for automatic detection of key phases for a contrast-enhanced scan, comprising:

• a memory encoding processor-executable routines;
• and
• a processing system comprising one or more processors and configured to access the memory and to execute the processor-executable routines, wherein the processor-executable routines, when executed by the processing system, cause the processing system to:
• acquire a series of images over time of a liver of a subject injected with a contrast bolus, wherein the series of images span multiple phases of the contrast-enhanced scan, and wherein the multiple phases comprise a pre-contrast phase, an aortic phase, a portal-venous phase, and a delayed phase;
• input the series of images into a trained neural network framework;
• utilize the trained neural network framework to separately segment an aorta and a portal vein from each image of the series of images;
• utilize the trained neural network framework to select exact respective time points for at least the aortic phase and the portal-venous phase based on the aorta and the portal vein segmented in each image of the series of images;
• and
• output a first set of images associated with the aortic phase and a second set of images associated with the portal-venous phase based on the exact respective time points selected for the aortic phase and the portal-venous phase.

A non-transitory computer-readable medium, the computer-readable medium comprising processor-executable code that when executed by a processing system comprising one or more processors, causes the processing system to:

• acquire a series of images over time of a liver of a subject injected with a contrast bolus, wherein the series of images span multiple phases of a contrast-enhanced scan, and wherein the multiple phases comprise a pre-contrast phase, an aortic phase, a portal-venous phase, and a delayed phase;
• input the series of images into a trained neural network framework;
• utilize the trained neural network framework to separately segment an aorta and a portal vein from each image of the series of images;
• utilize the trained neural network framework to select exact respective time points for at least the aortic phase and the portal-venous phase based on the aorta and the portal vein segmented in each image of the series of images;
• and
• output a first set of images associated with the aortic phase and a second set of images associated with the portal-venous phase based on the exact respective time points selected for the aortic phase and the portal-venous phase.