End-to-End Co-Optimization of Adaptive $k$-space Sampling and Reconstruction for Dynamic MRI

George Yiasemis, Jan-Jakob Sonke, Jonas Teuwen

Keywords: Adaptive MRI Sampling, Dynamic MRI Reconstruction, Joint Optimization

TL;DR: We introduce an end-to-end method that adaptively chooses case-specific k-space samples for dynamic MRI and reconstructs them jointly, improving image quality over fixed and non-adaptive sampling strategies, especially at higher acceleration factors.

Abstract: Accelerating dynamic MRI is essential for advancing clinical imaging and improving patient comfort. Most deep learning methods for dynamic MRI reconstruction rely on predetermined or random subsampling patterns that are uniformly applied across all temporal frames. Such strategies ignore temporal correlations and fail to optimize sampling for individual cases. To address this, we propose E2E-ADS-Recon, an end-to-end framework for adaptive dynamic MRI subsampling and reconstruction. The framework integrates an Adaptive Dynamic Sampler (ADS), which generates case-specific sampling patterns for a given acceleration factor, with a dynamic MRI reconstruction network that reconstructs the adaptively sampled data into a dynamic image sequence. The ADS can produce either frame-specific or unified patterns across time frames. We evaluate the method on multi-coil cardiac cine MRI data under both 1D and 2D sampling settings and compare it with standard and optimized non-adaptive baselines. E2E-ADS-Recon achieves superior reconstruction quality, particularly at higher acceleration rates. These results highlight the benefit of case-specific adaptive sampling and demonstrate the potential of joint sampling–reconstruction optimization for dynamic MRI. Code and trained models will be made publicly available upon acceptance.

Primary Subject Area: Image Acquisition and Reconstruction

Secondary Subject Area: Application: Radiology

Registration Requirement: Yes

Reproducibility: https://github.com/NKI-AI/direct/tree/adpt

Visa & Travel: No

Read CFP & Author Instructions: Yes

Originality Policy: Yes

Single-blind & Not Under Review Elsewhere: Yes

LLM Policy: Yes

Midl Latex Submission Checklist: Ensure no LaTeX errors during compilation., Replace NNN with your OpenReview submission ID., Includes \documentclass{midl}, \jmlryear{2026}, \jmlrworkshop, \jmlrvolume, \editors, and correct \bibliography command., Did not override options of the hyperref package., Did not use the times package., Use the correct spelling and format, avoid Unicode characters, and use LaTeX equivalents instead., Any math in the title and abstract must be enclosed within $...$., Did not override the bibliography style defined in midl.cls and did not use \begin{thebibliography} directly to insert references., Avoid using \scalebox; use \resizebox when needed., Included all necessary figures and removed *unused* files in the zip archive., Removed special formatting, visual annotations, and highlights used during rebuttal., All special characters in the paper and .bib file use LaTeX commands (e.g., \'e for é)., No separate supplementary PDF uploads., Acknowledgements, references, and appendix must start after the main content.

Submission Number: 102