Dual-domain self-supervised learning for accelerated non-Cartesian MRI reconstruction

While enabling accelerated acquisition and improved reconstruction accuracy, current deep MRI reconstruction networks are typically supervised, require fully sampled data, and are limited to Cartesian sampling patterns. These factors limit their practical adoption as fully-sampled MRI is prohibitively time-consuming to acquire clinically. Further, non-Cartesian sampling patterns are particularly desirable as they are more amenable to acceleration and show improved motion robustness. To this end, we present a fully self-supervised approach for accelerated non-Cartesian MRI reconstruction which leverages self-supervision in both k-space and image domains. In training, the undersampled data are split into disjoint k-space domain partitions. For the k-space self-supervision, we train a network to reconstruct the input undersampled data from both the disjoint partitions and from itself. For the image-level self-supervision, we enforce appearance consistency obtained from the original undersampled data and the two partitions. Experimental results on our simulated multi-coil non-Cartesian MRI dataset demonstrate that DDSS can generate high-quality reconstruction that approaches the accuracy of the fully supervised reconstruction, outperforming previous baseline methods. Finally, DDSS is shown to scale to highly challenging real-world clinical MRI reconstruction acquired on a portable low-field (0.064 T) MRI scanner with no data available for supervised training while demonstrating improved image quality as compared to traditional reconstruction, as determined by a radiologist study.     

Accelerating Cartesian MRI by domain-transform manifold learning in phase-encoding direction

This study developed a domain-transform framework comprising domain-transform manifold learning with an initial analytic transform to accelerate Cartesian magnetic resonance imaging (DOTA-MRI). The proposed method directly transforms undersampled Cartesian k-space data into a reconstructed image. In Cartesian undersampling, the k-space is fully or zero sampled in the data-acquisition direction (i.e., the frequency-encoding direction or the x-direction); one-dimensional (1D) inverse Fourier transform (IFT) along the x-direction on the undersampled k-space does not induce any aliasing. To exploit this, the algorithm first applies an analytic x-direction 1D IFT to the undersampled Cartesian k-space input, and subsequently transforms it into a reconstructed image using deep neural networks. The initial analytic transform (i.e., 1D IFT) allows the fully connected layers of the neural network to learn 1D global transform only in the phase-encoding direction (i.e., the y-direction) instead of 2D transform. This drastically reduces the number of parameters to be learned from O(N²) to O(N) compared with the existing manifold learning algorithm (i.e., automated transform by manifold approximation) (AUTOMAP). This enables DOTA-MRI to be applied to high-resolution MR datasets, which has previously proved difficult to implement in AUTOMAP because of the enormous memory requirements involved. After the initial analytic transform, the manifold learning phase uses a symmetric network architecture comprising three types of layers: front-end convolutional layers, fully connected layers for the 1D global transform, and back-end convolutional layers. The front-end convolutional layers take 1D IFT of the undersampled k-space (i.e., undersampled data in the intermediate domain or in the ky-x domain) as input and performs data-domain restoration. The following fully connected layers learn the 1D global transform between the ky-x domain and the image domain (i.e., the y-x domain). Finally, the back-end convolutional layers reconstruct the final image by denoising in the image domain. DOTA-MRI exhibited superior performance over nine other existing algorithms, including state-of-the-art deep learning-based algorithms. The generality of the algorithm was demonstrated by experiments conducted under various sampling ratios, datasets, and noise levels.     

卷积神经网络重建欠采的磁共振图像

目的使用卷积神经网络(convolutional neural network,CNN)从欠采样的磁共振成像K空间数据快速重建出无伪影的高质量图像。材料与方法实验数据包含60位自愿者矢状位、横断位、冠状位全采的T1加权脑部MR图像,使用旋转和拉伸等操作对训练数据进行扩增。不同欠采模式的MR图像和金标准图像分别输入CNN进行训练,学习获得的网络可实现由欠采图像到全采集图像之间的非线性映射。重建时,将CNN重建图像的K空间与原始的K空间数据进行合并保真。实验中利用金标准图像,计算重建图像的峰值信噪比(peak signal to noise ratio,PSNR)、结构相似度(structural similarity,SSIM)和高频误差范数(high frequency error norm,HFEN),定量评价重建结果。结果 (1)CNN重建出的中央采样MR图像的PSNR、SSIM、HFEN分别为31.13、0.93、223.81,优于Tukey窗函数的25.69、0.86、482.75;(2)CNN重建出的伪随机采样MR图像的PSNR、SSIM、HFEN分别为32.78、0.95、195.51,优于压缩感知的31.01、0.93、184.69。结论卷积神经网络可以从欠采数据重建出高质量的磁共振图像,无论是主观的视觉效果还是客观的评价参数都优于传统的处理方法。与K空间中央连续采集相比,伪随机采样模式更有利于CNN重建出高质量的MR图像。     

Adaptive convolutional neural networks for accelerating magnetic resonance imaging via k-space data interpolation

Deep learning in k-space has demonstrated great potential for image reconstruction from undersampled k-space data in fast magnetic resonance imaging (MRI). However, existing deep learning-based image reconstruction methods typically apply weight-sharing convolutional neural networks (CNNs) to k-space data without taking into consideration the k-space data's spatial frequency properties, leading to ineffective learning of the image reconstruction models. Moreover, complementary information of spatially adjacent slices is often ignored in existing deep learning methods. To overcome such limitations, we have developed a deep learning algorithm, referred to as adaptive convolutional neural networks for k-space data interpolation (ACNN-k-Space), which adopts a residual Encoder-Decoder network architecture to interpolate the undersampled k-space data by integrating spatially contiguous slices as multi-channel input, along with k-space data from multiple coils if available. The network is enhanced by self-attention layers to adaptively focus on k-space data at different spatial frequencies and channels. We have evaluated our method on two public datasets and compared it with state-of-the-art existing methods. Ablation studies and experimental results demonstrate that our method effectively reconstructs images from undersampled k-space data and achieves significantly better image reconstruction performance than current state-of-the-art techniques. Source code of the method is available at https://gitlab.com/qgpmztmf/acnn-k-space.     

Deep model-based magnetic resonance parameter mapping network (DOPAMINE) for fast T1 mapping using variable flip angle method

Quantitative tissue characteristics, which provide valuable diagnostic information, can be represented by magnetic resonance (MR) parameter maps using magnetic resonance imaging (MRI); however, a long scan time is necessary to acquire them, which prevents the application of quantitative MR parameter mapping to real clinical protocols. For fast MR parameter mapping, we propose a deep model-based MR parameter mapping network called DOPAMINE that combines a deep learning network with a model-based method to reconstruct MR parameter maps from undersampled multi-channel k-space data. DOPAMINE consists of two networks: 1) an MR parameter mapping network that uses a deep convolutional neural network (CNN) that estimates initial parameter maps from undersampled k-space data (CNN-based mapping), and 2) a reconstruction network that removes aliasing artifacts in the parameter maps with a deep CNN (CNN-based reconstruction) and an interleaved data consistency layer by an embedded MR model-based optimization procedure. We demonstrated the performance of DOPAMINE in brain $T_1$ map reconstruction with a variable flip angle (VFA) model. To evaluate the performance of DOPAMINE, we compared it with conventional parallel imaging, low-rank based reconstruction, model-based reconstruction, and state-of-the-art deep-learning-based mapping methods for three different reduction factors ($R$ = 3, 5, and 7) and two different sampling patterns (1D Cartesian and 2D Poisson-disk). Quantitative metrics indicated that DOPAMINE outperformed other methods in reconstructing $T_1$ maps for all sampling patterns and reduction factors. DOPAMINE exhibited quantitatively and qualitatively superior performance to that of conventional methods in reconstructing MR parameter maps from undersampled multi-channel k-space data. The proposed method can thus reduce the scan time of quantitative MR parameter mapping that uses a VFA model.     

A wavelet-based regularized reconstruction algorithm for SENSE parallel MRI with applications to neuroimaging

To reduce scanning time and/or improve spatial/temporal resolution in some Magnetic Resonance Imaging (MRI) applications, parallel MRI acquisition techniques with multiple coils acquisition have emerged since the early 1990s as powerful imaging methods that allow a faster acquisition process. In these techniques, the full FOV image has to be reconstructed from the resulting acquired undersampled k-space data. To this end, several reconstruction techniques have been proposed such as the widely-used SENSitivity Encoding (SENSE) method. However, the reconstructed image generally presents artifacts when perturbations occur in both the measured data and the estimated coil sensitivity profiles. In this paper, we aim at achieving accurate image reconstruction under degraded experimental conditions (low magnetic field and high reduction factor), in which neither the SENSE method nor the Tikhonov regularization in the image domain give convincing results. To this end, we present a novel method for SENSE-based reconstruction which proceeds with regularization in the complex wavelet domain by promoting sparsity. The proposed approach relies on a fast algorithm that enables the minimization of regularized non-differentiable criteria including more general penalties than a classical ?(1) term. To further enhance the reconstructed image quality, local convex constraints are added to the regularization process. In vivo human brain experiments carried out on Gradient-Echo (GRE) anatomical and Echo Planar Imaging (EPI) functional MRI data at 1.5T indicate that our algorithm provides reconstructed images with reduced artifacts for high reduction factors.     

Adaptive diffusion priors for accelerated MRI reconstruction

Deep MRI reconstruction is commonly performed with conditional models that de-alias undersampled acquisitions to recover images consistent with fully-sampled data. Since conditional models are trained with knowledge of the imaging operator, they can show poor generalization across variable operators. Unconditional models instead learn generative image priors decoupled from the operator to improve reliability against domain shifts related to the imaging operator. Recent diffusion models are particularly promising given their high sample fidelity. Nevertheless, inference with a static image prior can perform suboptimally. Here we propose the first adaptive diffusion prior for MRI reconstruction, AdaDiff, to improve performance and reliability against domain shifts. AdaDiff leverages an efficient diffusion prior trained via adversarial mapping over large reverse diffusion steps. A two-phase reconstruction is executed following training: a rapid-diffusion phase that produces an initial reconstruction with the trained prior, and an adaptation phase that further refines the result by updating the prior to minimize data-consistency loss. Demonstrations on multi-contrast brain MRI clearly indicate that AdaDiff outperforms competing conditional and unconditional methods under domain shifts, and achieves superior or on par within-domain performance.     

膝关节前交叉韧带的MRI三维成像研究

目的 探讨MRI三维成像技术(M3D/cube T2WI)在膝关节前交叉韧带(anterior cruciate ligament,ACL)正常显示和损伤诊断的价值.方法 选取30例正常自愿者(对照组)及11例ACL损伤病例(损伤组),在GE 1.5 T磁共振分别行MRI常规膝关节矢状面T1WI、FSE压脂T2WI(f...     

Application of k-space energy spectrum analysis to susceptibility field mapping and distortion correction in gradient-echo EPI

Echo-planar imaging (EPI) is widely used in functional MRI studies. It is well known that EPI quality is usually degraded by geometric distortions, when there exist susceptibility field inhomogeneities. EPI distortions may be corrected if the field maps are available. It is possible to estimate the susceptibility field gradients from the phase reconstruction of a single-TE EPI image, after a successful phase-unwrapping procedure. However, in regions affected by pronounced field gradients, the phase-unwrapping of a single-TE image may fail, and therefore the estimated field maps may be incorrect. It has been reported that the field inhomogeneity may be calculated more reliably from T2*-weighted images corresponding to multiple TEs. However, the multi-TE MRI field mapping increases the scan time. Furthermore, the measured field maps may be invalid if the subject's position changes during dynamic scans. To overcome the limitations in conventional field mapping approaches, a novel k-space energy spectrum analysis algorithm is developed, which quantifies the spatially dependent echo-shifting effect and the susceptibility field gradients directly from the k-space data of single-TE gradient-echo EPI. Using the k-space energy spectrum analysis, susceptibility field gradients can be reliably measured without phase-unwrapping, and EPI distortions can be corrected without extra field mapping scans or pulse sequence modification. The reported technique can be used to retrospectively improve the image quality of the previously acquired EPI and functional MRI data, provided that the complex-domain k-space data are still available.     

Synthesized 7T MRI from 3T MRI via deep learning in spatial and wavelet domains

Ultra-high field 7T MRI scanners, while producing images with exceptional anatomical details, are cost prohibitive and hence highly inaccessible. In this paper, we introduce a novel deep learning network that fuses complementary information from spatial and wavelet domains to synthesize 7T T1-weighted images from their 3T counterparts. Our deep learning network leverages wavelet transformation to facilitate effective multi-scale reconstruction, taking into account both low-frequency tissue contrast and high-frequency anatomical details. Our network utilizes a novel wavelet-based affine transformation (WAT) layer, which modulates feature maps from the spatial domain with information from the wavelet domain. Extensive experimental results demonstrate the capability of the proposed method in synthesizing high-quality 7T images with better tissue contrast and greater details, outperforming state-of-the-art methods.     

Probabilistic 3D MRI atlas of the human cerebellar dentate/interposed nuclei

In a previous study, a three-dimensional (3D) MRI atlas of the human cerebellar nuclei was introduced based on findings in one healthy human subject [Dimitrova, A., Weber, J., Redies, C., Kindsvater, K., Maschke, M., Kolb, F.P., Forsting, M., Diener, H.C., Timmann, D., 2002. MRI atlas of the human cerebellar nuclei. NeuroImage 17, 240-255]. The present MRI investigation was designed to study variability of the anatomy of the dentate/interposed nuclei in a larger group of healthy subjects. Similar to our previous study, iron-induced susceptibility artifacts were used to visualize the cerebellar nuclei as hypointensities on MR images. Data of 63 healthy subjects (27 female, 36 male; mean age 45.3+/-13.4 years, age range 22--71 years) were included. A 3D axial volume of the cerebellum was acquired using a T2*-weighted FLASH sequence on a Siemens Sonata 1.5 T MR scanner. Each volume was registered, re-sampled to 1.00 x 1.00 x 1.00 mm(3) voxel size and spatially normalized into a standard proportional stereotaxic space using SPM99. Dentate/interposed nuclei were traced on axial images and saved as regions of interest using MRIcro-software by two independent examiners. A probabilistic 3D MRI atlas of the cerebellar dentate/interposed nuclei is presented based on findings in all subjects.     

Real-time cardiovascular magnetic resonance at 1.5?T using balanced SSFP and 40?ms resolution

Background

While cardiovascular magnetic resonance (CMR) commonly employs ECG-synchronized cine acquisitions with balanced steady-state free precession (SSFP) contrast at 1.5 T, recent developments at 3 T demonstrate significant potential for T1-weighted real-time imaging at high spatiotemporal resolution using undersampled radial FLASH. The purpose of this work was to combine both ideas and to evaluate a corresponding real-time CMR method at 1.5 T with SSFP contrast.

Methods

Radial gradient-echo sequences with fully balanced gradients and at least 15-fold undersampling were implemented on two CMR systems with different gradient performance. Image reconstruction by regularized nonlinear inversion (NLINV) was performed offline and resulted in real-time SSFP CMR images at a nominal resolution of 1.8 mm and with acquisition times of 40 ms.

Results

Studies of healthy subjects demonstrated technical feasibility in terms of robustness and general image quality. Clinical applicability with access to quantitative evaluations (e.g., ejection fraction) was confirmed by preliminary applications to 27 patients with typical indications for CMR including arrhythmias and abnormal wall motion. Real-time image quality was slightly lower than for cine SSFP recordings, but considered diagnostic in all cases.

Conclusions

Extending conventional cine approaches, real-time radial SSFP CMR with NLINV reconstruction provides access to individual cardiac cycles and allows for studies of patients with irregular heartbeat.     

Direct parametric reconstruction from undersampled (k,t)-space data in dynamic contrast enhanced MRI

The Magnetic Resonance Imaging (MRI) signal can be made sensitive to functional parameters that provide information about tissues. In dynamic contrast enhanced (DCE) MRI these functional parameters are related to the microvasculature environment and the concentration changes that occur rapidly after the injection of a contrast agent. Typically DCE images are reconstructed individually and kinetic parameters are estimated by fitting a pharmacokinetic model to the time-enhancement response; these methods can be denoted as “indirect”. If undersampling is present to accelerate the acquisition, techniques such as kt-FOCUSS can be employed in the reconstruction step to avoid image degradation. This paper suggests a Bayesian inference framework to estimate functional parameters directly from the measurements at high temporal resolution. The current implementation estimates pharmacokinetic parameters (related to the extended Tofts model) from undersampled (k, t)-space DCE MRI. The proposed scheme is evaluated on a simulated abdominal DCE phantom and prostate DCE data, for fully sampled, 4 and 8-fold undersampled (k, t)-space data. Direct kinetic parameters demonstrate better correspondence (up to 70% higher mutual information) to the ground truth kinetic parameters (of the simulated abdominal DCE phantom) than the ones derived from the indirect methods. For the prostate DCE data, direct kinetic parameters depict the morphology of the tumour better. To examine the impact on cancer diagnosis, a peripheral zone prostate cancer diagnostic model was employed to calculate a probability map for each method.     

Automated 3D reconstruction of the fetal thorax in the standard atlas space from motion-corrupted MRI stacks for 21–36 weeks GA range

Slice-to-volume registration (SVR) methods allow reconstruction of high-resolution 3D images from multiple motion-corrupted stacks. SVR-based pipelines have been increasingly used for motion correction for T2-weighted structural fetal MRI since they allow more informed and detailed diagnosis of brain and body anomalies including congenital heart defects (Lloyd et al., 2019). Recently, fully automated rigid SVR reconstruction of the fetal brain in the atlas space was achieved in Salehi et al. (2019) that used convolutional neural networks (CNNs) for segmentation and pose estimation. However, these CNN-based methods have not yet been applied to the fetal trunk region. Meanwhile, the existing rigid and deformable SVR (DSVR) solutions (Uus et al., 2020) for the fetal trunk region are limited by the requirement of manual input as well the narrow capture range of the classical gradient descent based registration methods that cannot resolve severe fetal motion frequently occurring at the early gestational age (GA). Furthermore, in our experience, the conventional 2D slice-wise CNN-based brain masking solutions are reportedly prone to errors that require manual corrections when applied on a wide range of acquisition protocols or abnormal cases in clinical setting.In this work, we propose a fully automated pipeline for reconstruction of the fetal thorax region for 21–36 weeks GA range T2-weighted MRI datasets. It includes 3D CNN-based intra-uterine localisation of the fetal trunk and landmark-guided pose estimation steps that allow automated DSVR reconstruction in the standard radiological space irrespective of the fetal trunk position or the regional stack coverage. The additional step for generation of the common template space and rejection of outliers provides the means for automated exclusion of stacks affected by low image quality or extreme motion. The pipeline was quantitatively evaluated on a series of experiments including fetal MRI datasets and simulated rotation motion. Furthermore, we performed a qualitative assessment of the image reconstruction quality in terms of the definition of vascular structures on 100 early (median 23.14 weeks) and late (median 31.79 weeks) GA group MRI datasets covering 21 to 36 weeks GA range.     

3D MRI诊断胎儿体表畸形的研究

目的 研究3D MRI显示胎儿体表正常结构和畸形的临床应用价值.方法 对34例孕妇行US检查和MRI检查.胎儿尸检及出生后随访证实体表畸形36例,共42处.MR扫描均采集单次激发快速自旋回波序列(SSFSE),厚层重T2WI,三维稳态进动快速成像(3D FIESTA)序列,然后对3D FIESTA序列原始数据行多平面重...     

原发性脑淋巴瘤的CT和MRI观察

目的:探讨原发性脑淋巴瘤的CT、MRI表现。材料与方法:回顾性分析25例未经治疗的原发性脑淋巴瘤的临床、病理及CT、MRI表现。结果:25例患者共计35个病灶,其中16例患者(19个病灶)同时作MRI检查,所有病灶CT平扫为等或稍高密度,T1WI为等、低信号,其中5个病灶伴局灶性高信号;12个病灶在T2WI上为等、低信号;14个病灶DWI为等、高信号。所有病灶均有增强,18个病灶为均匀增强。13例病理检查,病灶表现为瘤细胞密集、高核浆比,病灶内出血坏死少见。结论:免疫功能正常状态原发性脑淋巴瘤的CT、MR表现有一定特征性,但鉴别诊断仍需仔细谨慎。     

Multiatlas Fusion with a Hybrid CT Number Correction Technique for Subject-Specific Pseudo-CT Estimation in the Context of MRI-Only Radiation Therapy

ObjectiveTo propose a hybrid multiatlas fusion and correction approach to estimate a pseudo–computed tomography (pCT) image from T2-weighted brain magnetic resonance (MR) images in the context of MRI-only radiotherapy.Materials and MethodsA set of eleven pairs of T2-weighted MR and CT brain images was included. Using leave-one-out cross-validation, atlas MR images were registered to the target MRI with multimetric, multiresolution deformable registration. The subsequent deformations were applied to the atlas CT images, producing uncorrected pCT images. Afterward, a three-dimensional hybrid CT number correction technique was used. This technique uses information about MR intensity, spatial location, and tissue label from segmented MR images with the fuzzy c-means algorithm and combines them in a weighted fashion to correct Hounsfield unit values of the uncorrected pCT images. The corrected pCT images were then fused into a final pCT image.ResultsThe proposed hybrid approach proved to be performant in correcting Hounsfield unit values in terms of qualitative and quantitative measures. Average correlation was 0.92 and 0.91 for the proposed approach by taking the mean and the median, respectively, compared with 0.86 for the uncorrected unfused version. Average values of dice similarity coefficient for bone were 0.68 and 0.72 for the fused corrected pCT images by taking the mean and the median, respectively, compared with 0.65 for the uncorrected unfused version indicating a significant bone estimation improvement.ConclusionA hybrid fusion and correction method is presented to estimate a pCT image from T2-weighted brain MR images.     

Magnetic resonance tomography pattern of a normal hip joint in a child and the respective pattern following trophic disturbances which resulted from treatment of developmental dislocation

MRI gives unique opportunities to visualize all structures of pediatric hip both in normal and pathologic conditions. As such it is a valuable supplement to other methods used so far in hip imaging, e.g., conventional radiography and ultrasonography. The aim of this study was to show MR pictures of normal pediatric hip and following avascular necrosis in developmental hip dislocation. The value of MRI and these conventional radiography were also compared. Magnetic resonance imaging appeared to be superior to radiography in assessment of subcapital growth plate and its damage in most of the hips. The most common presentation of the area of subcapital growth plate of the proximal femur on MR images was that of hypointense band on T1-weighted images and hyperintense band on sequences using fat saturation. T1-weighted images were best for visualization of bony bridges containing yellow marrow crossing the growth plate. The sequences using fat saturation were good for visualization af any damage to the growth plate without possibility to distinguish fibrous or osseous changes. T1-weighted images together with FLASH 3D FAT SAT seem to be sufficient for the analysis of the growth plate and its damage.     

MRI differential diagnosis of intrahepatic biloma from subacute hematoma

The magnetic resonance imaging (MRI) differential diagnosis of intrahepatic biloma from intrahepatic subacute hematoma has been reported in two cases. The biloma was heterogenously intense on T1-weighted images and homogenously hyperintense on T2-weighted images. The hematoma was hyperintense on the both T1-and T2-weighted MR images. The clinical significance of this MRI difference is that intrahepatic biloma needs drainage, whereas intrahepatic hematoma can heal spontaneously.     

3．0T MR肝脏三维容积超快速多期动态增强扫描在肝局灶性病变诊断中的价值

目的：探讨肝脏三维容积超快速多期动态增强扫描（3D CE LAVA）对肝占位性病变的诊断价值。资料与方法：对76例经病理证实的肝脏肿瘤患者行三维容积超快速多期动态增强扫描和2D GRE T1WI增强扫描，并采用最大强度投影（MIP）和MPR方式进行图像重建，统计病变显示率及其强化程度。结果：全肝动脉期多时相三维动态增强MR扫描及其图像重建，可同时显示肝脏肿瘤的动态强化过程和肝血管形态，对肝脏肿瘤的诊断及鉴别诊断提供依据。结论：MRI 3D LAVA多期动态增强扫描无论在病灶的显示及定性诊断方面，还是在病灶血管及肝血管解剖的显示上均较2D GRE T1WI增强扫描具有较高的临床价值。