@article{lyu_convolutional_2026,

title = {Convolutional recurrent U‐net for cardiac cine MRI reconstruction via effective spatio‐temporal feature exploitation},

author = {D. Lyu and M. Staring and M. J. P. Van Osch and M. Doneva and H. J. Lamb and N. Pezzotti},

url = {https://aapm.onlinelibrary.wiley.com/doi/10.1002/mp.70245},

doi = {10.1002/mp.70245},

issn = {0094-2405, 2473-4209},

year  = {2026},

date = {2026-07-01},

volume = {53},

number = {1},

pages = {e70245},

abstract = {Abstract 

 Background 

 Cardiac Cine Magnetic Resonance Imaging (MRI) provides dynamic visualization of the heart's structure and function but is hindered by slow acquisition, requiring repeated breath‐holds that challenge sick patients. Accelerated imaging can mitigate these issues but potentially reduce spatial and temporal resolution. Therefore, innovative approaches are essential to ensure effective performance under high acceleration conditions. Deep learning‐based reconstruction methods show promise in enhancing image quality from highly undersampled data, accelerating scans while maintaining diagnostic accuracy. However, they often fail to effectively exploit the spatio‐temporal features inherent to cine MRI, which are essential for accurate reconstruction, thereby leaving room for further improvement. 

 Purpose 

 We aim to more effectively exploit the spatio‐temporal features inherent in cine MRI sequences by integrating convolutional recurrent operations with a U‐Net architecture, enhancing the reconstruction performance of cine MRI. 

 Methods 

 We developed a new deep learning model called 

 CRUNet‐MR 

 that enhances the extraction of spatio‐temporal features by combining convolutional recurrent operations with a U‐Net structure. This design ensures continuous extraction of temporal features while fusing fine‐grained spatial details with high‐level semantic information. Furthermore, dilated convolutions are incorporated to expand the spatial receptive field, and appropriate combinations of dilation factors are explored to further enhance overall performance. 

 Results 

 Training, validation, and testing were performed on the public CMRxRecon2023 dataset, using two views and four acceleration factors ranging from 4 to 24 with the given Auto‐Calibration Signal (ACS) area. The dataset consists of 120 subjects for training, 60 for validation, and 120 for testing. In general, the proposed CRUNet‐MR shows statistically significant differences with benchmark models and consistently outperforms them, particularly showcasing better reconstruction quality in dynamic regions, highlighting its effective extraction of spatio‐temporal features. Ablation studies further validated the design choices of CRUNet‐MR. The model demonstrated strong reconstruction performance, achieving an average SSIM of 0.986 at an acceleration factor of 4 and 0.971 at a factor of 8 across both views. Furthermore, CRUNet‐MR was validated on a small in‐house LUMC dataset, showing its generalization capability and rapid adaptability through fine‐tuning. 

 Conclusions 

 The proposed CRUNet‐MR model is well‐suited for cine MRI reconstruction, effectively leveraging spatio‐temporal features to reconstruct high‐quality images, especially in dynamic cardiac regions. This capability highlights its potential to support higher acceleration factors, enabling faster and more patient‐friendly cardiac imaging.},

keywords = {},

pubstate = {published},

tppubtype = {article}

}

@article{beljaards_deepdisorder_2026,

title = {DEEP‐DISORDER: Motion Correction in 3D MRI via Segment Reconstruction and Registration},

author = {L. Beljaards and M. Nagtegaal and C. Rao and Y. Dong and M. J. P. van Osch and N. Pezzotti and M. Doneva and M. Staring},

url = {https://analyticalsciencejournals.onlinelibrary.wiley.com/doi/10.1002/nbm.70286},

doi = {10.1002/nbm.70286},

issn = {0952-3480, 1099-1492},

year  = {2026},

date = {2026-07-01},

volume = {39},

number = {5},

pages = {e70286},

abstract = {ABSTRACT 

 3D MR image acquisition is inherently time intensive, rendering it susceptible to patient motion during scanning. This may introduce significant blurring and artifacts, potentially necessitating reacquisition. We propose a modular framework to retrospectively correct for intrascan motion in 3D brain MRI, without active motion tracking. Serving as the backbone of our approach is an existing distributed and incoherent sampling scheme (DISORDER), combined with a fast network trained for highly undersampled reconstruction. This enables approximate reconstructions of anatomy after every few seconds, using only a tiny fraction of k‐space data (< 2%). While these reconstructions are only approximate, we postulate they are sufficient to estimate motion patterns at said temporal resolution. Groupwise registration, notable for its elimination of registration bias, is utilized for estimating rigid motion parameters, which are leveraged to reconstruct the measured data with reduced motion artifacts. The approach was evaluated on 94 retrospectively and 3 prospectively motion‐corrupted in vivo 3D T1‐weighted brain MRI acquisitions. The estimated motion parameters matched the known retrospective motion with 0.06 mm and 0.13° accuracy, resulting in an improvement in reconstruction quality from to SSIM for the retrospective scans. The prospective scans improved from to SSIM after correction in the case of gradual motion and from to SSIM for extreme motion. In conclusion, the proposed approach, that is free of external tracking devices or navigators, successfully estimated and corrected 3D motion between small subportions of a scan. This resulted in vastly improved image quality, making volumetric MRI substantially more tolerant to motion.},

keywords = {},

pubstate = {published},

tppubtype = {article}

}

@article{rao_plugandplay_2026,

title = {A plug-and-play method for guided multi-contrast MRI reconstruction based on content/style modeling},

author = {C. Rao and M. Van Osch and N. Pezzotti and J. De Bresser and M. Van Buchem and L. Beljaards and J. Meineke and E. De Weerdt and H. Lu and M. Doneva and M. Staring},

url = {https://linkinghub.elsevier.com/retrieve/pii/S136184152600229X},

doi = {10.1016/j.media.2026.104160},

issn = {13618415},

year  = {2026},

date = {2026-07-01},

volume = {113},

pages = {104160},

keywords = {},

pubstate = {published},

tppubtype = {article}

}

@article{lyu_mcpmedsam_2025,

title = {MCP-MedSAM: A Powerful Lightweight Medical Segment Anything Model Trained with a Single GPU in Just One Day},

author = {D. Lyu and R. Gao and M. Staring},

url = {https://melba-journal.org/2025:008},

doi = {10.59275/j.melba.2025-4849},

issn = {2766-905X},

year  = {2025},

date = {2025-05-12},

urldate = {2025-05-12},

volume = {3},

issue = {May 2025},

pages = {135–151},

abstract = {Medical image segmentation involves partitioning medical images into meaningful regions, with a focus on identifying anatomical structures and lesions. It has broad applications in healthcare, and deep learning methods have enabled significant advancements in automating this process. Recently, the introduction of the Segmentation Anything Model (SAM), the first foundation model for segmentation task, has prompted researchers to adapt it for the medical domain to improve performance across various tasks. However, SAM’s large model size and high GPU requirements hinder its scalability and development in the medical domain. To address these challenges, research has increasingly focused on lightweight adaptations of SAM to reduce its parameter count, enabling training with limited GPU resources while maintaining competitive segmentation performance. In this work, we propose MCP-MedSAM, a powerful and lightweight medical SAM model designed to be trainable on a single A100 GPU with 40GB of memory within one day while delivering superior segmentation performance. Recognizing the significant internal differences between modalities and the need for direct segmentation target information within bounding boxes, we introduce two kinds of prompts: the modality prompt and the content prompt. After passing through the prompt encoder, their embedding representations can further improve the segmentation performance by incorporating more relevant information without adding significant training overhead. Additionally, we adopt an effective modality-based data sampling strategy to address data imbalance between modalities, ensuring more balanced performance across all modalities. Our method was trained and evaluated using a large-scale challenge dataset, compared to top-ranking methods on the challenge leaderboard, MCP-MedSAM achieved superior performance while requiring only one day of training on a single GPU. The code is publicly available at <a href='https://github.com/dong845/MCP-MedSAM'>https://github.com/dong845/MCP-MedSAM</a>},

keywords = {},

pubstate = {published},

tppubtype = {article}

}

@article{beljaards_aibased_2024,

title = {AI‐based motion artifact severity estimation in undersampled MRI allowing for selection of appropriate reconstruction models},

author = {L. Beljaards and N. Pezzotti and C. Rao and M. Doneva and M. J. P. Van Osch and M. Staring},

url = {https://aapm.onlinelibrary.wiley.com/doi/10.1002/mp.16918},

doi = {10.1002/mp.16918},

issn = {0094-2405, 2473-4209},

year  = {2024},

date = {2024-07-01},

volume = {51},

number = {5},

pages = {3555–3565},

abstract = {Abstract 

 Background 

 Magnetic Resonance acquisition is a time consuming process, making it susceptible to patient motion during scanning. Even motion in the order of a millimeter can introduce severe blurring and ghosting artifacts, potentially necessitating re‐acquisition. Magnetic Resonance Imaging (MRI) can be accelerated by acquiring only a fraction of k‐space, combined with advanced reconstruction techniques leveraging coil sensitivity profiles and prior knowledge. Artificial intelligence (AI)‐based reconstruction techniques have recently been popularized, but generally assume an ideal setting without intra‐scan motion. 

 Purpose 

 To retrospectively detect and quantify the severity of motion artifacts in undersampled MRI data. This may prove valuable as a safety mechanism for AI‐based approaches, provide useful information to the reconstruction method, or prompt for re‐acquisition while the patient is still in the scanner. 

 Methods 

 We developed a deep learning approach that detects and quantifies motion artifacts in undersampled brain MRI. We demonstrate that synthetically motion‐corrupted data can be leveraged to train the convolutional neural network (CNN)‐based motion artifact estimator, generalizing well to real‐world data. Additionally, we leverage the motion artifact estimator by using it as a selector for a motion‐robust reconstruction model in case a considerable amount of motion was detected, and a high data consistency model otherwise. 

 Results 

 Training and validation were performed on 4387 and 1304 synthetically motion‐corrupted images and their uncorrupted counterparts, respectively. Testing was performed on undersampled in vivo motion‐corrupted data from 28 volunteers, where our model distinguished head motion from motion‐free scans with 91% and 96% accuracy when trained on synthetic and on real data, respectively. It predicted a manually defined quality label (‘Good’, ‘Medium’ or ‘Bad’ quality) correctly in 76% and 85% of the time when trained on synthetic and real data, respectively. When used as a selector it selected the appropriate reconstruction network 93% of the time, achieving near optimal SSIM values. 

 Conclusions 

 The proposed method quantified motion artifact severity in undersampled MRI data with high accuracy, enabling real‐time motion artifact detection that can help improve the safety and quality of AI‐based reconstructions.},

keywords = {},

pubstate = {published},

tppubtype = {article}

}