Search papers, labs, and topics across Lattice.
This paper introduces a multi-dimensional primitive-based framework for dynamic contrast-enhanced MRI reconstruction that effectively disentangles anatomical structures, dynamic contrast enhancement, and residual motion into distinct temporal basis functions. By leveraging Gaussian and Gabor primitives, the method achieves competitive performance with conventional reconstruction techniques, demonstrating high-quality spatiotemporal reconstructions even at high undersampling rates. The architecture not only enhances reconstruction quality but also improves the accuracy of extracted enhancement curves for critical anatomical regions like the aorta and kidney.
Achieving high-quality MRI reconstructions at unprecedented undersampling rates by disentangling dynamic contrast and anatomy could revolutionize clinical imaging practices.
Reliable quantitative analysis of dynamic contrast-enhanced MRI requires high-quality spatiotemporal reconstructions at high undersampling rates. Scan-specific reconstructions using Gaussian and Gabor primitives have shown promising results without the need for large training datasets, but have not addressed the additional dimension of dynamic contrast. We propose a multi-dimensional, primitive based framework for dynamic contrast-enhanced MRI reconstruction that disentangles the underlying anatomy, the dynamic contrast enhancement, and residual motion into separate temporal basis functions, thereby enabling a geometrical interpretation of the representation. We show that this architecture achieves performance competitive with conventional reconstruction methods, both in reconstruction quality and in the accuracy of extracted aorta and kidney enhancement curves. The modular tier design extends naturally to additional dynamic factors and higher acceleration rates. Code available at https://github.com/compai-lab/ 2026-GaborDCE-spieker.
