Calibration-free DCE-MRI with Sub-second Temporal Resolution Using Interpretable Implicit Neural Representation

Motivation: Dynamic Contrast-Enhanced MRI (DCE-MRI) requires high temporal resolution to effectively capture motion and contrast dynamics. Current methods rely on pre-estimating temporal information from calibration data, limiting imaging performance. Goal(s): We aim to develop a novel reconstruction method for achieving sub-second temporal resolution in DCE-MRI without the need for calibration data. Approach: We introduce an unsupervised Implicit Neural Representation (INR) framework, where spatial coordinates and learnable temporal latent variables are used to model and reconstruct the DCE-MRI data directly. Results: Our method reconstructs artifact-free images, with even one spoke per frame, and the learned latent variables accurately capture both contrast changes and respiratory motion. Impact: The proposed method paves the way for artifact-free, high-temporal-resolution DCE-MRI for clinical applications. Moreover, the proposed INR framework makes calibrationless MRI reconstruction feasible and interpretable, offering a powerful tool for a wide range of imaging challenges.