Joint Reconstruction of Multiple Images and Motion in MRI: Application to Free-Breathing Myocardial T₂Quantification

Motion-Compensated MR CINE Reconstruction With Reconstruction-Driven Motion Estimation

In cardiac CINE, motion-compensated MR reconstruction (MCMR) is an effective approach to address highly undersampled acquisitions by incorporating motion information between frames. In this work, we propose a novel perspective for addressing the MCMR problem and a more integrated and efficient solution to the MCMR field. Contrary to state-of-the-art (SOTA) MCMR methods which break the original problem into two sub-optimization problems, i.e. motion estimation and reconstruction, we formulate this problem as a single entity with one single optimization. Our approach is unique in that the motion estimation is directly driven by the ultimate goal, reconstruction, but not by the canonical motion-warping loss (similarity measurement between motion-warped images and target images). We align the objectives of motion estimation and reconstruction, eliminating the drawbacks of artifacts-affected motion estimation and therefore error-propagated reconstruction. Further, we can deliver high-quality reconstruction and realistic motion without applying any regularization/smoothness loss terms, circumventing the non-trivial weighting factor tuning. We evaluate our method on two datasets: 1) an in-house acquired 2D CINE dataset for the retrospective study and 2) the public OCMR cardiac dataset for the prospective study. The conducted experiments indicate that the proposed MCMR framework can deliver artifact-free motion estimation and high-quality MR images even for imaging accelerations up to 20x, outperforming SOTA non-MCMR and MCMR methods in both qualitative and quantitative evaluation across all experiments. The code is available at https://github.com/JZPeterPan/MCMR-Recon-Driven-Motion.

Unrolled and rapid motion-compensated reconstruction for cardiac CINE MRI

In recent years Motion-Compensated MR reconstruction (MCMR) has emerged as a promising approach for cardiac MR (CMR) imaging reconstruction. MCMR estimates cardiac motion and incorporates this information in the reconstruction. However, two obstacles prevent the practical use of MCMR in clinical situations: First, inaccurate motion estimation often leads to inferior CMR reconstruction results. Second, the motion estimation frequently leads to a long processing time for the reconstruction. In this work, we propose a learning-based and unrolled MCMR framework that can perform precise and rapid CMR reconstruction. We achieve accurate reconstruction by developing a joint optimization between the motion estimation and reconstruction, in which a deep learning-based motion estimation framework is unrolled within an iterative optimization procedure. With progressive iterations, a mutually beneficial interaction can be established in which the reconstruction quality is improved with more accurate motion estimation. Further, we propose a groupwise motion estimation framework to speed up the MCMR process. A registration template based on the cardiac sequence average is introduced, while the motion estimation is conducted between the cardiac frames and the template. By applying this framework, cardiac sequence registration can be accomplished with linear time complexity. Experiments on 43 in-house acquired 2D CINE datasets indicate that the proposed unrolled MCMR framework can deliver artifacts-free motion estimation and high-quality CMR reconstruction even for imaging acceleration rates up to 20x. We compare our approach with state-of-the-art reconstruction methods and it outperforms them quantitatively and qualitatively in all adapted metrics across all acceleration rates.

Feasibility of online non-rigid motion correction for high-resolution supine breast MRI

PURPOSE

Conventional breast MRI is performed in the prone position with a dedicated coil. This allows high-resolution images without breast motion, but the patient position is inconsistent with that of other breast imaging modalities or interventions. Supine breast MRI may be an interesting alternative, but respiratory motion becomes an issue. Motion correction methods have typically been performed offline, for instance, the corrected images were not directly accessible from the scanner console. In this work, we seek to show the feasibility of a fast, online, motion-corrected reconstruction integrated into the clinical workflow.

METHODS

Fully sampled T2 -weighted (T2 w) and accelerated T1 -weighted (T1 w) breast supine MR images were acquired during free-breathing and were reconstructed using a non-rigid motion correction technique (generalized reconstruction by inversion of coupled systems). Online reconstruction was implemented using a dedicated system combining the MR raw data and respiratory signals from an external motion sensor. Reconstruction parameters were optimized on a parallel computing platform, and image quality was assessed by objective metrics and by radiologist scoring.

RESULTS

Online reconstruction time was 2 to 2.5 min. The metrics and the scores related to the motion artifacts significantly improved for both T2 w and T1 w sequences. The overall quality of T2 w images was approaching that of the prone images, whereas the quality of T1 w images remained significantly lower.

CONCLUSION

The proposed online algorithm allows a noticeable reduction of motion artifacts and an improvement of the diagnostic quality for supine breast imaging with a clinically acceptable reconstruction time. These findings serve as a starting point for further development aimed at improving the quality of T1 w images.

Deformation-Compensated Learning for Image Reconstruction Without Ground Truth

Deep neural networks for medical image reconstruction are traditionally trained using high-quality ground-truth images as training targets. Recent work on Noise2Noise (N2N) has shown the potential of using multiple noisy measurements of the same object as an alternative to having a ground-truth. However, existing N2N-based methods are not suitable for learning from the measurements of an object undergoing nonrigid deformation. This paper addresses this issue by proposing the deformation-compensated learning (DeCoLearn) method for training deep reconstruction networks by compensating for object deformations. A key component of DeCoLearn is a deep registration module, which is jointly trained with the deep reconstruction network without any ground-truth supervision. We validate DeCoLearn on both simulated and experimentally collected magnetic resonance imaging (MRI) data and show that it significantly improves imaging quality.

Comparison of T2 Quantification Strategies in the Abdominal-Pelvic Region for Clinical Use

OBJECTIVES

The aim of the study was to compare different magnetic resonance imaging (MRI) acquisition strategies appropriate for T2 quantification in the abdominal-pelvic area. The different techniques targeted in the study were chosen according to 2 main considerations: performing T2 measurement in an acceptable time for clinical use and preventing/correcting respiratory motion.

MATERIALS AND METHODS

Acquisitions were performed at 3 T. To select sequences for in vivo measurements, a phantom experiment was conducted, for which the T2 values obtained with the different techniques of interest were compared with the criterion standard (single-echo SE sequence, multiple acquisitions with varying echo time). Repeatability and temporal reproducibility studies for the different techniques were also conducted on the phantom. Finally, an in vivo study was conducted on 12 volunteers to compare the techniques that offer acceptable acquisition time for clinical use and either address or correct respiratory motion.

RESULTS

For the phantom study, the DESS and T2-preparation techniques presented the lowest precision (ρ2 = 0.9504 and ρ2 = 0.9849 respectively), and showed a poor repeatability/reproducibility compared with the other techniques. The strategy relying on SE-EPI showed the best precision and accuracy (ρ2 = 0.9994 and Cb = 0.9995). GRAPPATINI exhibited a very good precision (ρ2 = 0.9984). For the technique relying on radial TSE, the precision was not as good as GRAPPATINI (ρ2 = 0.9872). The in vivo study demonstrated good respiratory motion management for all of the selected techniques. It also showed that T2 estimate ranges were different from one method to another. For GRAPPATINI and radial TSE techniques, there were significant differences between all the different types of organs of interest.

CONCLUSIONS

To perform T2 measurement in the abdominal-pelvic region, one should favor a technique with acceptable acquisition time for clinical use, with proper respiratory motion management, with good repeatability, reproducibility, and precision. In this study, the techniques relying respectively on SE-EPI, radial TSE, and GRAPPATINI appeared as good candidates.

Endogenous assessment of myocardial injury with single-shot model-based non-rigid motion-corrected T1 rho mapping

BACKGROUND

Cardiovascular magnetic resonance T1ρ mapping may detect myocardial injuries without exogenous contrast agent. However, multiple co-registered acquisitions are required, and the lack of robust motion correction limits its clinical translation. We introduce a single breath-hold myocardial T1ρ mapping method that includes model-based non-rigid motion correction.

METHODS

A single-shot electrocardiogram (ECG)-triggered balanced steady state free precession (bSSFP) 2D adiabatic T1ρ mapping sequence that collects five T1ρ-weighted (T1ρw) images with different spin lock times within a single breath-hold is proposed. To address the problem of residual respiratory motion, a unified optimization framework consisting of a joint T1ρ fitting and model-based non-rigid motion correction algorithm, insensitive to contrast change, was implemented inline for fast (~ 30 s) and direct visualization of T1ρ maps. The proposed reconstruction was optimized on an ex vivo human heart placed on a motion-controlled platform. The technique was then tested in 8 healthy subjects and validated in 30 patients with suspected myocardial injury on a 1.5T CMR scanner. The Dice similarity coefficient (DSC) and maximum perpendicular distance (MPD) were used to quantify motion and evaluate motion correction. The quality of T1ρ maps was scored. In patients, T1ρ mapping was compared to cine imaging, T2 mapping and conventional post-contrast 2D late gadolinium enhancement (LGE). T1ρ values were assessed in remote and injured areas, using LGE as reference.

RESULTS

Despite breath holds, respiratory motion throughout T1ρw images was much larger in patients than in healthy subjects (5.1 ± 2.7 mm vs. 0.5 ± 0.4 mm, P < 0.01). In patients, the model-based non-rigid motion correction improved the alignment of T1ρw images, with higher DSC (87.7 ± 5.3% vs. 82.2 ± 7.5%, P < 0.01), and lower MPD (3.5 ± 1.9 mm vs. 5.1 ± 2.7 mm, P < 0.01). This resulted in significantly improved quality of the T1ρ maps (3.6 ± 0.6 vs. 2.1 ± 0.9, P < 0.01). Using this approach, T1ρ mapping could be used to identify LGE in patients with 93% sensitivity and 89% specificity. T1ρ values in injured (LGE positive) areas were significantly higher than in the remote myocardium (68.4 ± 7.9 ms vs. 48.8 ± 6.5 ms, P < 0.01).

CONCLUSIONS

The proposed motion-corrected T1ρ mapping framework enables a quantitative characterization of myocardial injuries with relatively low sensitivity to respiratory motion. This technique may be a robust and contrast-free adjunct to LGE for gaining new insight into myocardial structural disorders.

Variational multi-task MRI reconstruction: Joint reconstruction, registration and super-resolution

Motion degradation is a central problem in Magnetic Resonance Imaging (MRI). This work addresses the problem of how to obtain higher quality, super-resolved motion-free reconstructions from highly undersampled MRI data. In this work, we present for the first time a variational multi-task framework that allows joining three relevant tasks in MRI: reconstruction, registration and super-resolution. Our framework takes a set of multiple undersampled MR acquisitions corrupted by motion into a novel multi-task optimisation model, which is composed of an L² fidelity term that allows sharing representation between tasks, super-resolution foundations and hyperelastic deformations to model biological tissue behaviors. We demonstrate that this combination yields significant improvements over sequential models and other bi-task methods. Our results exhibit fine details and compensate for motion producing sharp and highly textured images compared to state of the art methods while keeping low CPU time. Our improvements are appraised on both clinical assessment and statistical analysis.

Compressed sensing plus motion (CS + M): A new perspective for improving undersampled MR image reconstruction

We address the problem of reconstructing high quality images from undersampled MRI data. This is a challenging task due to the highly ill-posed nature of the problem. In particular, in dynamic MRI scans, the interaction between the target structure and the physical motion affects the acquired measurements leading to blurring artefacts and loss of fine details. In this work, we propose a framework for dynamic MRI reconstruction framed under a new multi-task optimisation model called Compressed Sensing Plus Motion (CS + M). Firstly, we propose a single optimisation problem that simultaneously computes the MRI reconstruction and the physical motion. Secondly, we show our model can be efficiently solved by breaking it up into two computationally tractable problems. The potentials and generalisation capabilities of our approach are demonstrated in different clinical applications including cardiac cine, cardiac perfusion and brain perfusion imaging. We show, through numerical experiments, that the proposed scheme reduces blurring artefacts, and preserves the target shape and fine details in the reconstruction. We also report the highest quality reconstruction under high undersampling rates in comparison to several state of the art techniques.

Cardiovascular magnetic resonance native T2 and T2* quantitative values for cardiomyopathies and heart transplantations: a systematic review and meta-analysis

BACKGROUND

The clinical application of cardiovascular magnetic resonance (CMR) T2 and T2* mapping is currently limited as ranges for healthy and cardiac diseases are poorly defined. In this meta-analysis we aimed to determine the weighted mean of T2 and T2* mapping values in patients with myocardial infarction (MI), heart transplantation, non-ischemic cardiomyopathies (NICM) and hypertension, and the standardized mean difference (SMD) of each population with healthy controls. Additionally, the variation of mapping outcomes between studies was investigated.

METHODS

The PRISMA guidelines were followed after literature searches on PubMed and Embase. Studies reporting CMR T2 or T2* values measured in patients were included. The SMD was calculated using a random effects model and a meta-regression analysis was performed for populations with sufficient published data.

RESULTS

One hundred fifty-four studies, including 13,804 patient and 4392 control measurements, were included. T2 values were higher in patients with MI, heart transplantation, sarcoidosis, systemic lupus erythematosus, amyloidosis, hypertrophic cardiomyopathy (HCM), dilated cardiomyopathy (DCM) and myocarditis (SMD of 2.17, 1.05, 0.87, 1.39, 1.62, 1.95, 1.90 and 1.33, respectively, P <  0.01) compared with controls. T2 values in iron overload patients (SMD = - 0.54, P = 0.30) and Anderson-Fabry disease patients (SMD = 0.52, P = 0.17) did both not differ from controls. T2* values were lower in patients with MI and iron overload (SMD of - 1.99 and - 2.39, respectively, P <  0.01) compared with controls. T2* values in HCM patients (SMD = - 0.61, P = 0.22), DCM patients (SMD = - 0.54, P = 0.06) and hypertension patients (SMD = - 1.46, P = 0.10) did not differ from controls. Multiple CMR acquisition and patient demographic factors were assessed as significant covariates, thereby influencing the mapping outcomes and causing variation between studies.

CONCLUSIONS

The clinical utility of T2 and T2* mapping to distinguish affected myocardium in patients with cardiomyopathies or heart transplantation from healthy myocardium seemed to be confirmed based on this meta-analysis. Nevertheless, variation of mapping values between studies complicates comparison with external values and therefore require local healthy reference values to clinically interpret quantitative values. Furthermore, disease differentiation seems limited, since changes in T2 and T2* values of most cardiomyopathies are similar.

Accelerated free-breathing whole-heart 3D T2 mapping with high isotropic resolution

Purpose

To enable free‐breathing whole‐heart 3D T₂ mapping with high isotropic resolution in a clinically feasible and predictable scan time. This 3D motion‐corrected undersampled signal matched (MUST) T₂ map is achieved by combining an undersampled motion‐compensated T₂‐prepared Cartesian acquisition with a high‐order patch‐based reconstruction.

Methods

The 3D MUST‐T₂ mapping acquisition consists of an electrocardiogram‐triggered, T₂‐prepared, balanced SSFP sequence with nonselective saturation pulses. Three undersampled T₂‐weighted volumes are acquired using a 3D Cartesian variable‐density sampling with increasing T₂ preparation times. A 2D image‐based navigator is used to correct for respiratory motion of the heart and allow 100% scan efficiency. Multicontrast high‐dimensionality undersampled patch‐based reconstruction is used in concert with dictionary matching to generate 3D T₂ maps. The proposed framework was evaluated in simulations, phantom experiments, and in vivo (10 healthy subjects, 2 patients) with 1.5‐mm³ isotropic resolution. Three‐dimensional MUST‐T₂ was compared against standard multi‐echo spin‐echo sequence (phantom) and conventional breath‐held single‐shot 2D SSFP T₂ mapping (in vivo).

Results

Three‐dimensional MUST‐T₂ showed high accuracy in phantom experiments (R² > 0.99). The precision of T₂ values was similar for 3D MUST‐T₂ and 2D balanced SSFP T₂ mapping in vivo (5 ± 1 ms versus 4 ± 2 ms, P = .52). Slightly longer T₂ values were observed with 3D MUST‐T₂ in comparison to 2D balanced SSFP T₂ mapping (50.7 ± 2 ms versus 48.2 ± 1 ms, P < .05). Preliminary results in patients demonstrated T₂ values in agreement with literature values.

Conclusion

The proposed approach enables free‐breathing whole‐heart 3D T₂ mapping with high isotropic resolution in about 8 minutes, achieving accurate and precise T₂ quantification of myocardial tissue in a clinically feasible scan time.

Motion-compensated reconstruction of magnetic resonance images from undersampled data

Magnetic resonance imaging of patients who find difficulty lying still or holding their breath can be challenging. Unresolved intra-frame motion yields blurring artifacts and limits spatial resolution. To correct for intra-frame non-rigid motion, such as in pediatric body imaging, this paper describes a multi-scale technique for joint estimation of the motion occurring during the acquisition and of the desired uncorrupted image. This technique regularizes the motion coefficients to enforce invertibility and minimize numerical instability. This multi-scale approach takes advantage of variable-density sampling patterns used in accelerated imaging to resolve large motion from a coarse scale. The resulting method improves image quality for a set of two-dimensional reconstructions from data simulated with independently generated deformations, with statistically significant increases in both peak signal to error ratio and structural similarity index. These improvements are consistent across varying undersampling factors and severities of motion and take advantage of the variable density sampling pattern.

Impact of denoising on precision and accuracy of saturation-recovery-based myocardial T1 mapping

PURPOSE

To evaluate the impact of a novel postprocessing denoising technique on accuracy and precision in myocardial T₁ mapping.

MATERIALS AND METHODS

This study introduces a fast and robust denoising method developed for magnetic resonance T₁ mapping. The technique imposes edge-preserving regularity and exploits the co-occurence of spatial gradients in the acquired T₁ -weighted images. The proposed approach was assessed in simulations, ex vivo data and in vivo imaging on a cohort of 16 healthy volunteers (12 males, average age 39 ± 8 years, 62 ± 9 bpm) both in pre- and postcontrast injection. The method was evaluated in myocardial T₁ mapping at 3T with a saturation-recovery technique that is accurate but sensitive to noise. ROIs in the myocardium and left-ventricle blood pool were analyzed by an experienced reader. Mean T₁ values and standard deviation were extracted and compared in all studies.

RESULTS

Simulations on synthetic phantom showed signal-to-noise ratio and sharpness improvement with the proposed method in comparison with conventional denoising. In vivo results demonstrated that our method preserves accuracy, as no difference in mean T₁ values was observed in the myocardium (precontrast: 1433/1426 msec, 95%CI: [-40.7, 55.9], p = 0.75, postcontrast: 766/759 msec, 95%CI: [-60.7, 77.2], p = 0.8). Meanwhile, precision was improved with standard deviations of T₁ values being significantly decreased (precontrast: 223/151 msec, 95%CI: [27.3, 116.5], p = 0.003, postcontrast: 176/135 msec, 95%CI: [5.5, 77.1], p = 0.03).

CONCLUSION

The proposed denoising method preserves accuracy and improves precision in myocardial T₁ mapping, with the potential to offer better map visualization and analysis.

LEVEL OF EVIDENCE

3 Technical Efficacy: Stage 1 J. Magn. Reson. Imaging 2017;46:1377-1388.