YDYiming Dong

Yiming Dong

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Yiming Dong has a total of 57 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Super-Resolution of Ultra-Low-Field MRI Using a GAN-Based Visual Transformer Network

    by Aram Salehi, Tavia E. Evans, Chloé Najac, Beatrice Lena, Yiming Dong, Ruben van Den Broek, Hieab H. H. Adams, Andrew Webb

  • Open Access

    Time‐Conditioned Zero‐Shot Self‐Supervised Reconstruction for Accelerated 3D Ultra‐Low‐Field MRI

    by Mart W. J. van Straten, Beatrice Lena, Chloé Najac, Ruben van Den Broek, Peter Börnert, Andrew Webb, Yiming Dong

    Purpose: Ultra‐low‐field (ULF) MRI provides a cost‐effective, portable imaging option but has relatively low SNR and long acquisition times compared to standard clinical scans. This study presents a time‐conditioned zero‐shot self‐supervised learning image reconstruction framework (ULF‐ZS‐SSL) to accelerate 3D‐acquired single‐coil ULF MRI without relying on external training data. In addition, for faster computation, a transfer‐learning (TL) variant (ULF‐ZS‐SSL‐TL) was implemented by pretraining on a small fully‐sampled ULF brain dataset and fine‐tuning on the target subject in a zero‐shot manner. Methods: This image reconstruction method combines a physics‐based data‐consistency step with a 3D residual network prior and sinusoidal time‐step embeddings to improve convergence speed. Data were acquired on a 47 mT Halbach‐based scanner using 3D turbo spin‐echo sequences with T 1 ‐, T 2 ‐, and inversion‐recovery–T 1 ‐weighted contrasts. Additional T 1 ‐weighted wrist scans were acquired to evaluate cross‐anatomy generalization. Both true and retrospectively undersampled data were compared with total variation (TV) and model‐based deep learning (MoDL). Results: The ULF‐ZS‐SSL method produced high‐quality reconstructions across all tested contrasts, outperforming zero‐filled and TV reconstructions, particularly at higher acceleration factors. Time‐step conditioning improved convergence speed, while ULF‐ZS‐SSL‐TL further accelerated the image reconstruction three‐fold, enabling full 3D reconstructions in about 3 min. Pretraining on brain data also worked well for wrist reconstructions, indicating cross‐anatomy generalization. Conclusion: The ULF‐ZS‐SSL framework enables accurate, training‐free reconstruction of undersampled single‐coil ULF MRI data, as does the ULF‐ZS‐SSL‐TL approach using minimal training data. The combination of physics‐based unrolling, time‐step conditioning, and transfer‐learning supports rapid and robust application in portable or resource‐limited ULF MRI systems.

  • Open Access

    4D Flow MRI at 0.6 T—Self-Gating Versus Camera-Based Respiratory Binning

    by Sébastien Emery, Luuk Jacobs, Jacob Malich, Gloria Wolkerstorfer, Yiming Dong, Ece Ercan, Jouke Smink, Martijn Nagtegaal, Sebastian Kozerke

    Four-dimensional (4D) flow MRI enables the comprehensive assessment of cardiovascular hemodynamics. To compensate for respiratory motion, self-gating strategies are typically used and perform reliably at clinical field strengths. With the recent push towards field strengths below 1 Tesla, these strategies need to be re-evaluated given the reduced signal-to-noise ratio (SNR). Camera-based, contactless respiratory monitoring offers an attractive alternative to self-gating, as it is unaffected by imaging. This study compared respiratory self-gating (SG) and camera-based (VE) binning for phase-contrast gradient-echo (PC-GRE) 4D flow MRI at 0.6 T. Data were acquired from twenty healthy subjects (age: 32.8 ± 12.6 years) using a pseudo-spiral undersampled Cartesian four-point velocity encoding scheme. Reconstructions were performed with FlowMRI-Net for the end-expiratory state using either SG or VE binning. SG and VE showed strong agreement, with cross-correlation coefficients of ~0.87, accuracies of ~0.87, and F1-scores of ~0.9. Velocity analysis revealed high concordance (R2 = 0.99; RMSE = 3.9 cm/s), with mean differences in peak velocity of 1.25 ± 2.36 cm/s. In this feasibility study, respiratory self-gating and camera-based binning yielded similar hemodynamic parameters from PC-GRE 4D flow MRI at 0.6 T, with the camera-based approach being independent of MR image SNR.


  • Robust Fat Suppression for High‐Resolution DWI at 5 T Using Slice‐Selection Gradient Modulation and Chemical Shift Encoding

    by Fan Liu, Yiming Dong, Wending Tang, Simin Liu, Shuo Chen, Guangqi Li, Diwei Shi, Xin Shao, Yuancheng Jiang, Huadan Xue, et al.

    Purpose: Chemical shift encoding (Dixon) can simultaneously remove multiple fat peaks in multi‐shot EPI diffusion‐weighted imaging (ms‐ EPI DWI ) at 3 T with low sensitivity to B 0 inhomogeneity. However, this method is not applicable at 5 T , since increased fat off‐resonance frequencies cause slice position mismatches between different fat peaks. In this work, we propose a two‐step strategy combining slice‐selection gradient modulation ( SSGM ) and Dixon to enhance fat suppression for ms‐ EPI DWI at 5 T. Methods: In the first step, SSGM adjusts the amplitudes of excitation and refocusing slice‐selection gradients according to the off‐resonance frequencies of methyl/methylene fat peaks, so that these fat slices are excited but not refocused. In the second step, the olefinic fat peak is chemical shift encoded and separated from the diffusion‐weighted water images through a joint water/fat separation algorithm with structured low‐rank regularization. The two‐step strategy was evaluated in the leg, head‐and‐neck, and prostate. Results: In vivo experiments demonstrated that the Dixon‐only methods cannot simultaneously suppress all fat peaks at 5 T , while this problem was addressed by combining SSGM and Dixon. SSGM showed superior suppression for methyl/methylene fat compared to SPAIR . The following Dixon further removed olefinic fat untouched by SPAIR . Qualitative analysis showed improved overall image quality for all anatomies. Prostate experiments showed that the proposed method is also applicable in reduced FOV acquisitions, high‐resolution (1.6‐mm isotropic) and high b value imaging (2800 s/mm 2 ). Conclusion: The proposed two‐step strategy improved fat suppression in ms‐ EPI DWI at 5 T , which can potentially enhance whole‐body disease screening and diagnosis.


  • Open Access

    Repeatability and reproducibility of rapid T1 mapping of brain tissues at 64 mT: A multicentre study

    by Beatrice Lena, Francesco Padormo, Rui Pedro A.G. Teixeira, Carly Bennallick, James Gholam, Ruben van Den Broek, Samson Lecurieux Lafayette, Irene Vavasour, Mara Cercignani, Derek K. Jones, et al.

    Very-low-field MRI (<100 mT) holds promise for Point-of-Care brain imaging applications, including stroke and multiple sclerosis, with T1 mapping emerging as a key biomarker for brain development and pathology. However, current low-field T1 mapping protocols suffer from long acquisition times and limited multi-site repeatability. This study aimed to improve T1 mapping at 64 mT using a clinically feasible 10-minute protocol and assess repeatability and reproducibility across sites. We present an analysis of the repeatability and reproducibility of rapid T1 measurements in a commercially available phantom and in 60 volunteers, scanned with a portable 64 mT MRI systems at six sites. T1 mapping was performed using an undersampled 3D inversion-recovery turbo spin-echo sequence with a 10.8-minute scan time, and reconstructed with a locally low-rank approach. Our results in phantom demonstrated high reproducibility in T1 measurements (below 3% differences from the average), with non-significant differences between sites. Longitudinal measurements demonstrated high repeatability over time both in vivo and in phantom settings in one site, with minimal variability (average Coefficient of Variation of 0.6%). Average in vivo T1 values for white matter and cortex were 290 ± 6 ms and 332 ± 8 ms, respectively and the values demonstrated high reproducibility, with differences of less than 4% from the average across sites. Our results demonstrate the feasibility of multi-site in vivo T1 mapping at 64 mT, providing normative T1 values at this field strength and supporting its use as a quantitative biomarker in clinical applications.


  • Open Access

    Navigator‐free multi‐shot diffusion MRI via non‐local low‐rank reconstruction

    by Yiming Dong, Xinyu Ye, Chang Li, Matthias J. P. van Osch, Peter Börnert

    Purpose: To develop a non‐local low‐rank (NLLR) reconstruction method for multi‐shot EPI (ms‐EPI) in DWI, addressing phase inconsistencies and noise issues while maintaining high spatial resolution in clinically feasible scan times. Theory and Methods: Single‐shot EPI (ss‐EPI) is widely used for DWI but suffers from geometric distortions and T 2 * blurring. ms‐EPI improves spatial resolution but introduces shot‐to‐shot phase variations requiring correction strategies. Traditional navigator‐based approaches may increase acquisition time. Recent low‐rank regularization reconstruction techniques, such as locally low‐rank (LLR) methods, can estimate the phase errors but rely strictly on local neighborhood information along the shot dimension. The proposed NLLR method extends this framework by leveraging non‐local patch matching by grouping similar image patches across spatially distant image locations, enhancing non‐local redundancy exploitation for improved phase estimation and correction as well as noise suppression. The method was validated in simulations and in vivo experiments and compared to existing post‐processing denoising and navigator‐free approaches. Results: In simulation experiments, compared to post‐processing denoising algorithms, NLLR demonstrated superior noise suppression and structural preservation across all metrics, even when reconstructing from a single diffusion direction. In the in‐vivo experiments, NLLR outperformed conventional navigator‐free approaches particularly regarding noise suppression. Fractional anisotropy maps reconstructed using NLLR exhibited improved visualization of fine structures with improved SNR, with performance differences becoming more pronounced at higher resolutions. Conclusion: The proposed NLLR approach provides an efficient and good solution for high‐resolution DWI reconstruction, improving image quality while mitigating phase variations and noise.


  • Open Access

    Rapid quantitative MRI at 46 mT : Accelerated T 1 and T 2 mapping with low‐rank reconstructions

    by Yiming Dong, Chloé Najac, Matthias J. P. van Osch, Andrew Webb, Peter Börnert, Beatrice Lena

    Purpose: To evaluate accelerated T 1 ‐ and T 2 ‐mapping techniques for ultra–low‐field MRI using low‐rank reconstruction methods. Methods: Two low‐rank–based algorithms, image‐based locally low‐rank (LLR) and k‐space–based structured low‐rank (SLR), were implemented to accelerate T 1 and T 2 mapping on a 46 mT Halbach MRI scanner. Data were acquired with 3D turbo spin‐echo sequences using variable‐density poisson‐disk random sampling patterns. For validation, phantom and in vivo experiments were performed on six healthy volunteers to compare the obtained values with literature and to study reconstruction performance at different undersampling factors and spatial resolutions. In addition, the reconstruction performance of the LLR and SLR algorithms for T 1 mapping was compared using retrospective undersampling datasets. Total scan times were reduced from 45/38 min (R = 1) to 23/19 min (R = 2) and 11/9 min (R = 4) for a 2.5 × 2.5 × 5 mm 3 resolution, and to 18/16 min (R = 4) for a higher in‐plane resolution 1.5 × 1.5 × 5 mm 3 for T 1 /T 2 mapping, respectively. Results: Both LLR and SLR algorithms successfully reconstructed T 1 and T 2 maps from undersampled data, significantly reducing scan times and eliminating undersampling artifacts. Phantom validation showed that consistent T 1 and T 2 values were obtained at different undersampling factors up to R = 4. For in vivo experiments, comparable image quality and estimated T 1 and T 2 values were obtained for fully sampled and undersampled (R = 4) reconstructions, both of which were in line with the literature values. Conclusions: The use of low‐rank reconstruction allows significant acceleration of T 1 and T 2 mapping in low‐field MRI while maintaining image quality.


  • Open Access

    Chemical shift–encoded multishot EPI for navigator‐free prostate DWI

    by Yiming Dong, David Atkinson, Kirsten Koolstra, Matthias J. P. van Osch, Peter Börnert

    Purpose: DWI is an important contrast for prostate MRI to enable early and accurate detection of cancer. This study introduces a Dixon 3‐shot‐EPI protocol with structured low‐rank reconstruction for navigator‐free DWI. The aim is to overcome the limitations of single‐shot EPI (ssh‐EPI), such as geometric distortions and fat signal interference, while addressing the motion‐induced phase variations of multishot EPI and simultaneously allowing water/fat separation.Methods: DWI data were acquired from 7 healthy volunteers using both Dixon 3‐shot EPI and standard fat‐suppressed ssh‐EPI with similar scan times for comparison. Two readers evaluated image quality using a 5‐point Likert scale regarding different aspects. The ADC values were quantitatively compared between protocols. To show feasibility in a clinical setting, the protocol was applied to two patients.Results: From the reader scores, Dixon 3‐shot EPI significantly reduced geometric distortion compared with ssh‐EPI (p < 0.01), with no significant differences in edge definition, SNR, or overall image quality. There was no significant difference in ADC values between the two protocols. However, the Dixon multishot‐EPI protocol offered advantages such as self‐referenced B0 map–driven distortion correction, greater flexibility in imaging parameters, and superior fat suppression. In the patient data, the lesion could be clearly identified in both protocols and on the associated ADC maps.Conclusion: The proposed Dixon 3‐shot‐EPI protocol shows promise as an alternative to ssh‐EPI for prostate DWI, providing reduced geometric distortions and improved fat suppression. It addresses common DWI issues based on EPI and enhances scanning flexibility, indicating potential for optimized imaging.


  • Open Access

    Structured low‐rank reconstruction for navigator‐free water/fat separated multi‐shot diffusion‐weightedEPI

    by Yiming Dong, Kirsten Koolstra, Ziyu Li, Malte Riedel, Matthias J. P. van Osch, Peter Börnert

    Purpose: Multi‐shot diffusion‐weighted EPI allows an increase in image resolution and reduced geometric distortions and can be combined with chemical‐shift encoding (Dixon) to separate water/fat signals. However, such approaches suffer from physiological motion‐induced shot‐to‐shot phase variations. In this work, a structured low‐rank‐based navigator‐free algorithm is proposed to address the challenge of simultaneously separating water/fat signals and correcting for physiological motion‐induced shot‐to‐shot phase variations in multi‐shot EPI‐based diffusion‐weighted MRI.Theory and Methods: We propose an iterative, model‐based reconstruction pipeline that applies structured low‐rank regularization to estimate and eliminate the shot‐to‐shot phase variations in a data‐driven way, while separating water/fat images. The algorithm is tested in different anatomies, including head–neck, knee, brain, and prostate. The performance is validated in simulations and in‐vivo experiments in comparison to existing approaches.Results: In‐vivo experiments and simulations demonstrated the effectiveness of the proposed algorithm compared to extra‐navigated and an alternative self‐navigation approach. The proposed algorithm demonstrates the capability to reconstruct in the multi‐shot/Dixon hybrid space domain under‐sampled datasets, using the same number of acquired EPI shots compared to conventional fat‐suppression techniques but eliminating fat signals through chemical‐shift encoding. In addition, partial Fourier reconstruction can also be achieved by using the concept of virtual conjugate coils in conjunction with the proposed algorithm.Conclusion: The proposed algorithm effectively eliminates the shot‐to‐shot phase variations and separates water/fat images, making it a promising solution for future DWI on different anatomies.


  • Open Access

    Water/fat separation for self‐navigated diffusion‐weighted multishot echo‐planar imaging

    by Yiming Dong, Malte Riedel, Kirsten Koolstra, Matthias J. P. van Osch, Peter Börnert

    The purpose of this study was to develop a self‐navigation strategy to improve scan efficiency and image quality of water/fat‐separated, diffusion‐weighted multishot echo‐planar imaging (ms‐EPI). This is accomplished by acquiring chemical shift‐encoded diffusion‐weighted data and using an appropriate water‐fat and diffusion‐encoded signal model to enable reconstruction directly from k‐space data. Multishot EPI provides reduced geometric distortion and improved signal‐to‐noise ratio in diffusion‐weighted imaging compared with single‐shot approaches. Multishot acquisitions require corrections for physiological motion‐induced shot‐to‐shot phase errors using either extra navigators or self‐navigation principles. In addition, proper fat suppression is important, especially in regions with large B0 inhomogeneity. This makes the use of chemical shift encoding attractive. However, when combined with ms‐EPI, shot‐to‐shot phase navigation can be challenging because of the spatial displacement of fat signals along the phase‐encoding direction. In this work, a new model‐based, self‐navigated water/fat separation reconstruction algorithm is proposed. Experiments in legs and in the head–neck region of 10 subjects were performed to validate the algorithm. The results are compared with an image‐based, two‐dimensional (2D) navigated water/fat separation approach for ms‐EPI and with a conventional fat saturation approach. Compared with the 2D navigated method, the use of self‐navigation reduced the shot duration time by 30%–35%. The proposed algorithm provided improved diffusion‐weighted water images in both leg and head–neck regions compared with the 2D navigator‐based approach. The proposed algorithm also produced better fat suppression compared with the conventional fat saturation technique in the B0 inhomogeneous regions. In conclusion, the proposed self‐navigated reconstruction algorithm can produce superior water‐only diffusion‐weighted EPI images with less artefacts compared with the existing methods.