Fig. 1

Overview of the deep learning framework for multi-vessel classification, segmentation, and phenomapping of phase-contrast MRI. The framework includes the Deep Classification + Segmentation (DCS) and Deep Temporal Clustering (DTC) models. DCS model: a 2D + time tunable UNet3 + with multi-class classification guidance. DTC model: a deep temporal clustering model architecture. MSE = Mean squared error, KL = Kullback–Leibler divergence, CGM = Classification-Guided Module adapted from UNet3 + 32.