Abstract: A neural network architecture and method for analysis of time series images from an image source employs a 3D-UNet convolutional neural network (CNN) configured to receive the time series images and generate spatiotemporal feature maps therefrom. Multiple sub-convolutional neural network output prongs based on an SRNet architecture receive the feature maps and simultaneously generate inferences for image segmentation, regression of values, and multi-landmark localization.

Abstract: A method and system for automated correction of phase error in MRI-based flow evaluation employs a computer processor programmed to execute a trained convolutional neural network (CNN) to receive and process image data comprising flow velocity data in three directions and magnitude data collected from a region of interest over a scan period from magnetic resonance imaging instrumentation. The image data is processed using the trained CNN to generate three output channels with pixelwise inferred corrections for the flow velocity data which are further smoothed using a regression algorithm. The smoothed corrections are added to the original image data to generate corrected flow data, which may be used for flow visualization and quantization.

Abstract: A method for improving resolution of a lower resolution image includes inputting at least one digital image into a convolutional neural network (CNN) comprising one of a SRNet and a UNet to output a two-dimensional image having a higher resolution. In some embodiments, multiple image frames may be processed using a 3D CNN to generate a two-dimensional output image.