HierAdaptMR: Cross-Center Cardiac MRI Reconstruction with Hierarchical Feature Adapters

Ruru Xu; Ilkay Oksuz

doi:10.1007/978-3-032-17734-6_28

HierAdaptMR: Cross-Center Cardiac MRI Reconstruction with Hierarchical Feature Adapters

Ruru Xu
, Ilkay Oksuz^*

^*Corresponding author for this work

Department of Computer Engineering

Istanbul Technical University

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

Abstract

Deep learning-based cardiac MRI reconstruction faces significant domain shift challenges when deployed across multiple clinical centers with heterogeneous scanner configurations and imaging protocols. We propose HierAdaptMR, a hierarchical feature adaptation framework that addresses multi-level domain variations through parameter-efficient adapters. Our method employs Protocol-Level Adapters for sequence-specific characteristics and Center-Level Adapters for scanner-dependent variations, built upon a variational unrolling backbone. A Universal Adapter enables generalization to entirely unseen centers through stochastic training that learns center-invariant adaptations. The framework utilizes multi-scale SSIM loss with frequency domain enhancement and contrast-adaptive weighting for robust optimization. Comprehensive evaluation on the CMRxRecon2025 dataset spanning 5+ centers, 10+ scanners, and 9 modalities demonstrates superior cross-center generalization while maintaining reconstruction quality.

Original language	English
Title of host publication	Statistical Atlases and Computational Models of the Heart. Regular and CMRxRecon Challenge Papers - 16th International Workshop, STACOM 2025, Held in Conjunction with MICCAI 2025, Revised Selected Papers
Editors	Oscar Camara, Esther Puyol Antón, Charlène Mauge, Alistair Young, Maxime Sermesant, Marta Varela, Yingliang Ma, Rasmus Paulsen, Chengyan Wang, Qian Tao
Publisher	Springer Science and Business Media Deutschland GmbH
Pages	299-310
Number of pages	12
ISBN (Print)	9783032177339
DOIs	https://doi.org/10.1007/978-3-032-17734-6_28
Publication status	Published - 2026
Event	16th International Workshop on Statistical Atlases and Computational Models of the Heart, STACOM 2025, Held in Conjunction with MICCAI 2025 - Daejeon, Korea, Republic of Duration: 27 Sept 2025 → 27 Sept 2025

Publication series

Name	Lecture Notes in Computer Science
Volume	16459 LNCS
ISSN (Print)	0302-9743
ISSN (Electronic)	1611-3349

Conference

Conference	16th International Workshop on Statistical Atlases and Computational Models of the Heart, STACOM 2025, Held in Conjunction with MICCAI 2025
Country/Territory	Korea, Republic of
City	Daejeon
Period	27/09/25 → 27/09/25

Bibliographical note

Publisher Copyright:
© The Author(s), under exclusive license to Springer Nature Switzerland AG 2026.

Keywords

Cardiac MRI
MRI Reconstruction

Access to Document

10.1007/978-3-032-17734-6_28

Cite this

Xu, R., & Oksuz, I. (2026). HierAdaptMR: Cross-Center Cardiac MRI Reconstruction with Hierarchical Feature Adapters. In O. Camara, E. Puyol Antón, C. Mauge, A. Young, M. Sermesant, M. Varela, Y. Ma, R. Paulsen, C. Wang, & Q. Tao (Eds.), Statistical Atlases and Computational Models of the Heart. Regular and CMRxRecon Challenge Papers - 16th International Workshop, STACOM 2025, Held in Conjunction with MICCAI 2025, Revised Selected Papers (pp. 299-310). (Lecture Notes in Computer Science; Vol. 16459 LNCS). Springer Science and Business Media Deutschland GmbH. https://doi.org/10.1007/978-3-032-17734-6_28

Xu, Ruru ; Oksuz, Ilkay. / HierAdaptMR : Cross-Center Cardiac MRI Reconstruction with Hierarchical Feature Adapters. Statistical Atlases and Computational Models of the Heart. Regular and CMRxRecon Challenge Papers - 16th International Workshop, STACOM 2025, Held in Conjunction with MICCAI 2025, Revised Selected Papers. editor / Oscar Camara ; Esther Puyol Antón ; Charlène Mauge ; Alistair Young ; Maxime Sermesant ; Marta Varela ; Yingliang Ma ; Rasmus Paulsen ; Chengyan Wang ; Qian Tao. Springer Science and Business Media Deutschland GmbH, 2026. pp. 299-310 (Lecture Notes in Computer Science).

@inproceedings{337f55e6d87d448d9b50200e3f7f6444,

title = "HierAdaptMR: Cross-Center Cardiac MRI Reconstruction with Hierarchical Feature Adapters",

abstract = "Deep learning-based cardiac MRI reconstruction faces significant domain shift challenges when deployed across multiple clinical centers with heterogeneous scanner configurations and imaging protocols. We propose HierAdaptMR, a hierarchical feature adaptation framework that addresses multi-level domain variations through parameter-efficient adapters. Our method employs Protocol-Level Adapters for sequence-specific characteristics and Center-Level Adapters for scanner-dependent variations, built upon a variational unrolling backbone. A Universal Adapter enables generalization to entirely unseen centers through stochastic training that learns center-invariant adaptations. The framework utilizes multi-scale SSIM loss with frequency domain enhancement and contrast-adaptive weighting for robust optimization. Comprehensive evaluation on the CMRxRecon2025 dataset spanning 5+ centers, 10+ scanners, and 9 modalities demonstrates superior cross-center generalization while maintaining reconstruction quality.",

keywords = "Cardiac MRI, MRI Reconstruction",

author = "Ruru Xu and Ilkay Oksuz",

note = "Publisher Copyright: {\textcopyright} The Author(s), under exclusive license to Springer Nature Switzerland AG 2026.; 16th International Workshop on Statistical Atlases and Computational Models of the Heart, STACOM 2025, Held in Conjunction with MICCAI 2025 ; Conference date: 27-09-2025 Through 27-09-2025",

year = "2026",

doi = "10.1007/978-3-032-17734-6\_28",

language = "English",

isbn = "9783032177339",

series = "Lecture Notes in Computer Science",

publisher = "Springer Science and Business Media Deutschland GmbH",

pages = "299--310",

editor = "Oscar Camara and \{Puyol Ant{\'o}n\}, Esther and Charl{\`e}ne Mauge and Alistair Young and Maxime Sermesant and Marta Varela and Yingliang Ma and Rasmus Paulsen and Chengyan Wang and Qian Tao",

booktitle = "Statistical Atlases and Computational Models of the Heart. Regular and CMRxRecon Challenge Papers - 16th International Workshop, STACOM 2025, Held in Conjunction with MICCAI 2025, Revised Selected Papers",

address = "Germany",

}

Xu, R & Oksuz, I 2026, HierAdaptMR: Cross-Center Cardiac MRI Reconstruction with Hierarchical Feature Adapters. in O Camara, E Puyol Antón, C Mauge, A Young, M Sermesant, M Varela, Y Ma, R Paulsen, C Wang & Q Tao (eds), Statistical Atlases and Computational Models of the Heart. Regular and CMRxRecon Challenge Papers - 16th International Workshop, STACOM 2025, Held in Conjunction with MICCAI 2025, Revised Selected Papers. Lecture Notes in Computer Science, vol. 16459 LNCS, Springer Science and Business Media Deutschland GmbH, pp. 299-310, 16th International Workshop on Statistical Atlases and Computational Models of the Heart, STACOM 2025, Held in Conjunction with MICCAI 2025, Daejeon, Korea, Republic of, 27/09/25. https://doi.org/10.1007/978-3-032-17734-6_28

HierAdaptMR: Cross-Center Cardiac MRI Reconstruction with Hierarchical Feature Adapters. / Xu, Ruru; Oksuz, Ilkay.
Statistical Atlases and Computational Models of the Heart. Regular and CMRxRecon Challenge Papers - 16th International Workshop, STACOM 2025, Held in Conjunction with MICCAI 2025, Revised Selected Papers. ed. / Oscar Camara; Esther Puyol Antón; Charlène Mauge; Alistair Young; Maxime Sermesant; Marta Varela; Yingliang Ma; Rasmus Paulsen; Chengyan Wang; Qian Tao. Springer Science and Business Media Deutschland GmbH, 2026. p. 299-310 (Lecture Notes in Computer Science; Vol. 16459 LNCS).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › peer-review

TY - GEN

T1 - HierAdaptMR

T2 - 16th International Workshop on Statistical Atlases and Computational Models of the Heart, STACOM 2025, Held in Conjunction with MICCAI 2025

AU - Xu, Ruru

AU - Oksuz, Ilkay

N1 - Publisher Copyright: © The Author(s), under exclusive license to Springer Nature Switzerland AG 2026.

PY - 2026

Y1 - 2026

N2 - Deep learning-based cardiac MRI reconstruction faces significant domain shift challenges when deployed across multiple clinical centers with heterogeneous scanner configurations and imaging protocols. We propose HierAdaptMR, a hierarchical feature adaptation framework that addresses multi-level domain variations through parameter-efficient adapters. Our method employs Protocol-Level Adapters for sequence-specific characteristics and Center-Level Adapters for scanner-dependent variations, built upon a variational unrolling backbone. A Universal Adapter enables generalization to entirely unseen centers through stochastic training that learns center-invariant adaptations. The framework utilizes multi-scale SSIM loss with frequency domain enhancement and contrast-adaptive weighting for robust optimization. Comprehensive evaluation on the CMRxRecon2025 dataset spanning 5+ centers, 10+ scanners, and 9 modalities demonstrates superior cross-center generalization while maintaining reconstruction quality.

AB - Deep learning-based cardiac MRI reconstruction faces significant domain shift challenges when deployed across multiple clinical centers with heterogeneous scanner configurations and imaging protocols. We propose HierAdaptMR, a hierarchical feature adaptation framework that addresses multi-level domain variations through parameter-efficient adapters. Our method employs Protocol-Level Adapters for sequence-specific characteristics and Center-Level Adapters for scanner-dependent variations, built upon a variational unrolling backbone. A Universal Adapter enables generalization to entirely unseen centers through stochastic training that learns center-invariant adaptations. The framework utilizes multi-scale SSIM loss with frequency domain enhancement and contrast-adaptive weighting for robust optimization. Comprehensive evaluation on the CMRxRecon2025 dataset spanning 5+ centers, 10+ scanners, and 9 modalities demonstrates superior cross-center generalization while maintaining reconstruction quality.

KW - Cardiac MRI

KW - MRI Reconstruction

UR - https://www.scopus.com/pages/publications/105040642355

U2 - 10.1007/978-3-032-17734-6_28

DO - 10.1007/978-3-032-17734-6_28

M3 - Conference contribution

AN - SCOPUS:105040642355

SN - 9783032177339

T3 - Lecture Notes in Computer Science

SP - 299

EP - 310

BT - Statistical Atlases and Computational Models of the Heart. Regular and CMRxRecon Challenge Papers - 16th International Workshop, STACOM 2025, Held in Conjunction with MICCAI 2025, Revised Selected Papers

A2 - Camara, Oscar

A2 - Puyol Antón, Esther

A2 - Mauge, Charlène

A2 - Young, Alistair

A2 - Sermesant, Maxime

A2 - Varela, Marta

A2 - Ma, Yingliang

A2 - Paulsen, Rasmus

A2 - Wang, Chengyan

A2 - Tao, Qian

PB - Springer Science and Business Media Deutschland GmbH

Y2 - 27 September 2025 through 27 September 2025

ER -

Xu R, Oksuz I. HierAdaptMR: Cross-Center Cardiac MRI Reconstruction with Hierarchical Feature Adapters. In Camara O, Puyol Antón E, Mauge C, Young A, Sermesant M, Varela M, Ma Y, Paulsen R, Wang C, Tao Q, editors, Statistical Atlases and Computational Models of the Heart. Regular and CMRxRecon Challenge Papers - 16th International Workshop, STACOM 2025, Held in Conjunction with MICCAI 2025, Revised Selected Papers. Springer Science and Business Media Deutschland GmbH. 2026. p. 299-310. (Lecture Notes in Computer Science). doi: 10.1007/978-3-032-17734-6_28