Generalized Autocalibrating Partially Parallel Acquisitions Research Articles

Accelerated cardiac cine with spatio-coil regularized deep learning reconstruction.

To develop an iterative deep learning (DL) reconstruction with spatio-coil regularization and multichannel k-space data consistency for accelerated cine imaging. This study proposes a Spatio-Coil Regularized DL (SCR-DL) approach for iterative deep learning reconstruction incorporating multicoil information in data consistency and regularizer. SCR-DL uses shift-invariant convolutional kernels to interpolate missing k-space lines and reconstruct individual coil images, followed by a regularizer that operates simultaneously across spatial and coil dimensions using learned image priors. At 8-fold acceleration, SCR-DL was compared with Generalized Autocalibrating Partially Parallel Acquisition (GRAPPA), sensitivity encoding (SENSE)-based DL and spatio-temporal regularized (STR)-DL reconstruction. In the retrospective undersampled cine, images were quantitatively evaluated using normalized mean square error (NMSE) and structural similarity index measure (SSIM). Additionally, agreement for left-ventricular ejection fraction and left-ventricular mass were assessed using prospectively accelerated cine images at 2-fold and 8-fold accelerations. The SCR-DL algorithm successfully reconstructed highly accelerated cine images. SCR-DL had significant improvements in NMSE (0.03 ± 0.02) and SSIM (91.4% ± 2.7%) compared with GRAPPA (NMSE: 0.09 ± 0.04, SSIM: 69.9% ± 11.1%; p < 0.001), SENSE-DL (NMSE: 0.07 ± 0.04, SSIM: 86.9% ± 3.2%; p < 0.001), and STR-DL (NMSE: 0.04 ± 0.03, SSIM: 90.0% ± 2.5%; p < 0.001) with retrospective undersampled cine. Despite the 3-fold reduction in scan time, there was no difference between left-ventricular ejection fraction (59.8 ± 4.5 vs. 60.8 ± 4.8, p = 0.46) or left-ventricular mass (73.6 ± 19.4 g vs. 73.2 ± 19.7 g, p = 0.95) between R = 2 and R = 8 prospectively accelerated cine images. SCR-DL enabled highly accelerated cardiac cine imaging, significantly reducing breath-hold time. Compared with GRAPPA or SENSE-DL, images reconstructed with SCR-DL showed superior NMSE and SSIM.

Bayesian merged utilization of GRAPPA and SENSE (BMUGS) for in-plane accelerated reconstruction increases fMRI detection power

In fMRI, capturing brain activity during a task is dependent on how quickly the k-space arrays for each volume image are obtained. Acquiring the full k-space arrays can take a considerable amount of time. Under-sampling k-space reduces the acquisition time, but results in aliased, or “folded,” images after applying the inverse Fourier transform (IFT). GeneRalized Autocalibrating Partial Parallel Acquisition (GRAPPA) and SENSitivity Encoding (SENSE) are parallel imaging techniques that yield reconstructed images from subsampled arrays of k-space. With GRAPPA operating in the spatial frequency domain and SENSE in image space, these techniques have been separate but can be merged to reconstruct the subsampled k-space arrays more accurately. Here, we propose a Bayesian approach to this merged model where prior distributions for the unknown parameters are assessed from a priori k-space arrays. The prior information is utilized to estimate the missing spatial frequency values, unalias the voxel values from the posterior distribution, and reconstruct into full field-of-view images. Our Bayesian technique successfully reconstructed simulated and experimental fMRI time series with no aliasing artifacts while decreasing temporal variation and increasing task detection power.

Accelerated High-Resolution Deep Learning Reconstruction Turbo Spin Echo MRI of the Knee at 7 T.

The aim of this study was to compare the image quality of 7 T turbo spin echo (TSE) knee images acquired with varying factors of parallel-imaging acceleration reconstructed with deep learning (DL)-based and conventional algorithms. This was a prospective single-center study. Twenty-three healthy volunteers underwent 7 T knee magnetic resonance imaging. Two-, 3-, and 4-fold accelerated high-resolution fat-signal-suppressing proton density (PD-fs) and T1-weighted coronal 2D TSE acquisitions with an encoded voxel volume of 0.31 × 0.31 × 1.5 mm3 were acquired. Each set of raw data was reconstructed with a DL-based and a conventional Generalized Autocalibrating Partially Parallel Acquisition (GRAPPA) algorithm. Three readers rated image contrast, sharpness, artifacts, noise, and overall quality. Friedman analysis of variance and the Wilcoxon signed rank test were used for comparison of image quality criteria. The mean age of the participants was 32.0 ± 8.1 years (15 male, 8 female). Acquisition times at 4-fold acceleration were 4 minutes 15 seconds (PD-fs, Supplemental Video is available at http://links.lww.com/RLI/A938) and 3 minutes 9 seconds (T1, Supplemental Video available at http://links.lww.com/RLI/A939). At 4-fold acceleration, image contrast, sharpness, noise, and overall quality of images reconstructed with the DL-based algorithm were significantly better rated than the corresponding GRAPPA reconstructions (P < 0.001). Four-fold accelerated DL-reconstructed images scored significantly better than 2- to 3-fold GRAPPA-reconstructed images with regards to image contrast, sharpness, noise, and overall quality (P ≤ 0.031). Image contrast of PD-fs images at 2-fold acceleration (P = 0.087), image noise of T1-weighted images at 2-fold acceleration (P = 0.180), and image artifacts for both sequences at 2- and 3-fold acceleration (P ≥ 0.102) of GRAPPA reconstructions were not rated differently than those of 4-fold accelerated DL-reconstructed images. Furthermore, no significant difference was observed for all image quality measures among 2-fold, 3-fold, and 4-fold accelerated DL reconstructions (P ≥ 0.082). This study explored the technical potential of DL-based image reconstruction in accelerated 2D TSE acquisitions of the knee at 7 T. DL reconstruction significantly improved a variety of image quality measures of high-resolution TSE images acquired with a 4-fold parallel-imaging acceleration compared with a conventional reconstruction algorithm.

Acquisition Acceleration of Ultra-low Field MRI with Parallel Imaging and Compressed Sensing in Microtesla Fields.

In recent years, ultra-low field (ULF) magnetic resonance imaging (MRI) has gained widespread attention due to its advantages, such as low cost, light weight, and portability. However, the low signal-to-noise ratio (SNR) leads to a long scan time. Herein, we study the acceleration performance of parallel imaging (PI) and compressed sensing (CS) in different kspace sampling strategies at 0.12 mT. This study employs phantoms to assess the efficiency of acceleration methods at ULF MRI, in which signals are detected by ultra-sensitive superconducting quantum interference devices (SQUIDs). We compare the performance of fast Fourier transform (FFT), generalized auto-calibrating partially parallel acquisitions (GRAPPA), and eigenvector-based SPIRiT (ESPIRiT) in Cartesian sampling, while also evaluating non-uniform FFT (NUFFT), GRAPPA operator gridding, and ESPIRiT in nonCartesian sampling. We design a resolution phantom to investigate the effectiveness of these methods in maintaining image resolution. In Cartesian sampling, GRAPPA and ESPIRiT jointly regularized by total variation and ℓ1-norm (TVJℓ1 -ESPIRiT) methods reconstructed good-quality phantom images with an acceleration factor of R = 2. In contrast, TVJℓ1-ESPIRiT exhibited improved image quality and much less signal loss even for R = 4. In radial sampling, TVJℓ1-ESPIRiT reduced the acquisition time to 1.69 minutes at R = 4, with a respective improvement of 12.26 dB in peak SNR compared to NUFFT. The resolution phantom imaging showed that the reconstructions by PI and CS maintained the original resolution of 2 mm. This study improves the practicality of ULF MRI at microtesla fields by implementing imaging acceleration with PI and CS in different k-space sampling.

In-vitro Detection of Intramammary-like Macrocalcifications Using Susceptibility-weighted MR Imaging Techniques at 1.5T.

The aim of our study was to investigate the technical accuracy of susceptibility-weighted imaging (SWI) and quantitative susceptibility mapping (QSM) created to detect intramammary-like calcifications depending on different TEs, volume, and type of calcification samples at 1.5T. Jello-embedded particles of blackboard chalk and ostrich eggshell ranging in size from 4 to 25 mm2 were used to simulate intramammary calcifications after testing different base substances and calcifications for their suitability to be used in breast phantoms. Breast phantoms were systematically examined using CT and an optimized 3D multi-echo gradient echo pulse sequence with following parameters: TR/TE, 22/1.88-15.52 ms in 1.24 ms increments; reconstructed voxel, 0.5 × 0.5 × 1.1 mm3; receiver bandwidth, 1120 Hz/Px; flip angle, 15°; integrated parallel imaging technique with a GeneRalized Autocalibrating Partial Parallel Acquisition (GRAPPA) factor of 2/24; and a total acquisition time of 3:00 min. A qualitative evaluation of the dependence of the visualization of calcification samples on volume and TE value was followed by a calculation of the SNR, the contrast-to-noise ratio (CNR) and the creation of SWI and QSM in the sense of a (semi)-quantitative analysis of the images. Jello proved to be a suitable base substance for preparing breast phantoms for SW MRI. Blackboard chalk and ostrich eggshell proved to be suitable for mimicking intramammary-like calcifications. The decrease in the median SNR of the blackboard chalk samples was significantly higher than the corresponding value of the ostrich eggshell samples over the entire TE range (47.5 to 17.0 vs. 16.0 to 6.56, P < 0.0001). The increase in the median CNR of the blackboard chalk samples was significantly higher than the corresponding value of the ostrich eggshell samples over the entire TE range (2.46 to 35.0 vs. 20.2 to 36.8, P = 0.007). With increasing TE value, the signal void volume of the calcification particle increases in the magnitude images as well as in SWI and QSM. Due to the blooming effect, the median gradients of the TE-based changes in signal void volumes were higher in SWI than in magnitude images and in QSM, regardless of the type of calcification particle examined. The maximum magnetic susceptibility of ostrich eggshell samples varied in a TE range of 1.88 to 15.52 ms from -7.2 to -2.51 ppm and that of blackboard chalk from -2.0 to -1.7 ppm. Compared to the manually measured volumes of the calcification particles, both MR-based measurements and CT examinations overestimated the actual sample size. The (non)-significant overestimation in the MRI-data is dependent on the set TE. The CT-based hyperdense volumes were overestimated compared to the corresponding manually measured sample volumes in a range of 109.8%-315.2% for ostrich eggshell samples (P = 0.016) and in a range of 39.9%-156.4% for blackboard chalk samples (P = 0.69). Our systematic in-vitro investigation of magnitude images, SWI, and QSM revealed that various set TE values, different volumes, and compositions of calcifications have a significant impact on visualizing intramammary(-like) calcifications.

Comparison between Conventional Breath-hold and Respiratory-triggered Magnetic Resonance Cholangiopancreatography with and without Compressed Sensing: Cross-sectional Study.

The application of compressed sensing (CS) has enabled breath-hold 3D-MRCP with a shorter acquisition time in clinical practice. To compare the image quality of breath-hold (BH) and respiratory-triggered (RT) 3D-MRCP with or without CS application in the same study population. In this retrospective study, from February to July 2020, a total of 98 consecutive patients underwent four different acquisition types of 3D-MRCP.; 1) BH MRCP with the generalized autocalibrating partially parallel acquisition (GRAPPA) (BH-GRAPPA), 2) RT-GRAPPA-MRCP, 3) RT-CS-MRCP and 4) BH-CS-MRCP. Relative contrast of common bile duct, 5-scale visibility score of biliary pancreatic ducts, 3-scale artifact score and 5-scale overall image quality score were evaluated by two abdominal radiologists. Relative contrast value was significantly higher in BH-CS or RT-CS than in RT-GRAPPA (0.90 ± 0.057 and 0.89 ± 0.079, respectively, vs. 0.82 ± 0.071, p < 0.01) or BH-GRAPPA (vs. 0.77 ± 0.080, p < 0.01). The area affected by artifact was significantly lower in BH-CS among 4 MRCPs (p < 0.08). Overall image quality score in BH-CS was significantly higher than BH-GRAPPA (3.40 vs. 2.71, p < 0.01). There were no significant differences between RT-GRAPPA and BH-CS (vs. 3.13, p = 0.67) in overall image quality. In this study, our results revealed BH-CS had higher relative contrast and comparable or superior image quality among four MRCP sequences.

Accelerated Cardiac MRI Cine with Use of Resolution Enhancement Generative Adversarial Inline Neural Network.

Background Cardiac cine can benefit from deep learning-based image reconstruction to reduce scan time and/or increase spatial and temporal resolution. Purpose To develop and evaluate a deep learning model that can be combined with parallel imaging or compressed sensing (CS). Materials and Methods The deep learning model was built on the enhanced super-resolution generative adversarial inline neural network, trained with use of retrospectively identified cine images and evaluated in participants prospectively enrolled from September 2021 to September 2022. The model was applied to breath-hold electrocardiography (ECG)-gated segmented and free-breathing real-time cine images collected with reduced spatial resolution with use of generalized autocalibrating partially parallel acquisitions (GRAPPA) or CS. The deep learning model subsequently restored spatial resolution. For comparison, GRAPPA-accelerated cine images were collected. Diagnostic quality and artifacts were evaluated by two readers with use of Likert scales and compared with use of Wilcoxon signed-rank tests. Agreement for left ventricle (LV) function, volume, and strain was assessed with Bland-Altman analysis. Results The deep learning model was trained on 1616 patients (mean age ± SD, 56 years ± 16; 920 men) and evaluated in 181 individuals, 126 patients (mean age, 57 years ± 16; 77 men) and 55 healthy subjects (mean age, 27 years ± 10; 15 men). In breath-hold ECG-gated segmented cine and free-breathing real-time cine, the deep learning model and GRAPPA showed similar diagnostic quality scores (2.9 vs 2.9, P = .41, deep learning vs GRAPPA) and artifact score (4.4 vs 4.3, P = .55, deep learning vs GRAPPA). Deep learning acquired more sections per breath-hold than GRAPPA (3.1 vs one section, P < .001). In free-breathing real-time cine, the deep learning showed a similar diagnostic quality score (2.9 vs 2.9, P = .21, deep learning vs GRAPPA) and lower artifact score (3.9 vs 4.3, P < .001, deep learning vs GRAPPA). For both sequences, the deep learning model showed excellent agreement for LV parameters, with near-zero mean differences and narrow limits of agreement compared with GRAPPA. Conclusion Deep learning-accelerated cardiac cine showed similarly accurate quantification of cardiac function, volume, and strain to a standardized parallel imaging method. © RSNA, 2023 Supplemental material is available for this article. See also the editorial by Vannier and Wang in this issue.

Accelerating GRAPPA reconstruction using SoC design for real-time cardiac MRI

Real-time cardiac MRI is a rapidly developing area of research that has the potential to improve the diagnosis and treatment of cardiovascular diseases. However, the acquisition of high-quality real-time cardiac MR (CMR) images is challenging as it requires a high frame rate and temporal resolution. To overcome this challenge, there have been recent efforts on several approaches including hardware-based improvements and image reconstruction techniques such as compressed sensing and parallel MRI. The use of parallel MRI techniques such as GRAPPA (Generalized Autocalibrating Partial Parallel Acquisition) is a promising approach for improving the temporal resolution of MRI and expanding its applications in clinical practice. However, the GRAPPA algorithm involves a significant amount of computation, particularly for high acceleration factors and large datasets. This can result in long reconstruction times, which can limit the ability to achieve real-time imaging or high frame rates. One solution to this challenge is to use specialized hardware i.e. field-programmable gate arrays (FPGAs). In this work, a novel 32-bit floating-point FPGA-based GRAPPA accelerator is proposed with an aim to reconstruct high-quality cardiac MR images at higher frame rates, making it well suited for real-time clinical applications. The proposed FPGA-based accelerator consists of custom-designed data processing units named as dedicated computational engines (DCEs) that allow for a continuous flow of data between the calibration and synthesis stages of GRAPPA reconstruction process. This greatly increases the throughput and reduces the latency of the overall proposed system. Moreover, a high-speed memory module (DDR4-SDRAM) is integrated with the proposed architecture to store the multi-coil MR data. An on-chip quad-core ARM Cortex-A53 processor is used to manage access control information required for data transfer between the DCEs and DDR4-SDRAM. The proposed accelerator is implemented on Xilinx Zynq UltraScale + MPSoC using high-level synthesis (HLS) and hardware descriptive language (HDL) with an aim to explore the trade-offs between the reconstruction time, resource utilization and design effort. Several experiments have been performed using in-vivo cardiac datasets i.e. 18-receiver coil and 30-receiver coil to evaluate the performance of the proposed accelerator. A comparison is performed with the contemporary CPU and GPU-based GRAPPA reconstruction methods in terms of reconstruction time, frames-per-second and reconstruction accuracy (RMSE and SNR). The results show that the proposed accelerator achieves speed-up factors up to 121× and 9× as compared to the contemporary CPU-based and GPU-based GRAPPA reconstruction methods, respectively. Moreover, it has been demonstrated that the proposed accelerator can achieve reconstruction rates of up to ∼27 frames-per-second while maintaining the visual quality of the reconstructed images.

High-efficiency 3D black-blood thoracic aorta imaging with patch-based low-rank tensor reconstruction.

Three-dimensional (3D) black-blood (BB) vessel wall imaging is a promising noninvasive imaging technique for assessing thoracic aortic diseases. We aimed to develop and evaluate a fast thoracic aorta vessel wall imaging method with patch-based low-rank tensor (Pt-LRT) reconstruction using the 3D-modulated variable flip angle fast-spin echo (vFA-FSE) sequence. The Pt-LRT technique adopts a low-rank tensor image model with regularization to explore the local low-rankness and nonlocal redundancies of the images to assess the thoracic aorta vessel wall. It uses high-order tensors to capture correlations between data in multiple dimensions and reconstructs images from highly undersampled data. For this study, 12 healthy participants and 2 patients with thoracic aortic diseases were evaluated at 3T magnetic resonance (MR). The reconstruction results were compared to the traditional generalized autocalibrating partially parallel acquisitions (GRAPPA) and ℓ1-SPIRiT reconstruction to assess the feasibility of the proposed framework. Quantitative analyses of the vessel wall thickness (VWT), internal diameter (ID), lumen area (LA), and contrast-to-noise ratio (CNR) between the lumen and vessel wall were performed on all healthy participants. Results demonstrated no significant differences between the GRAPPA and the proposed Pt-LRT in VWT, ID, or LA of the aorta (P<0.05). A higher mean CNR was attained with 3D patch-based low-rank tensor reconstruction than with ℓ1-SPIRiT reconstruction (49.4±10.8 vs. 38.9±8.2). The proposed 3D BB thoracic aorta vessel wall imaging method can reduce the scan time and produce an image quality that is in good agreement with the conventional GRAPPA acquisition, which takes approximately more than 8 min. This study also shows that the proposed Pt-LRT method substantially improves the visualization and sharpness of the vessel wall and the definition of the tissue boundary compared to the imaging obtained with ℓ1-SPIRiT.

Improved Parallel Magnetic Resonance Imaging Reconstruction with Sampling based on Normal Distribution

Generalized autocalibrating partially parallel acquisitions (GRAPPA) is a k-space-based magnetic resonance imaging (MRI) reconstruction algorithm that obtains coefficients by acquiring k-space center data with the Nyquist frequency requirement, fitting these data as auto-calibration signal (ACS). The under-sampled k-space data are then filled using the linear correlation of the neighboring points of the k-space. At high reduction factors, images reconstructed by using GRAPPA have high levels of noise even when the number of ACS lines is significantly increased. NL-GRAPPA (Non-linear GRAPPA) can improve GRAPPA for a comparatively impressive image quality. However, it is extremely difficult to reduce the sampling time significantly because of the high number of ACS lines that must be sampled. In this paper, we proposed an undersampling parallel MRI (pMRI) reconstruction technique based on the ideology of normal distribution to improve the traditional GRAPPA and NL-GRAPPA methods. The proposed method mainly takes advantage of the fact that the data in the center of the k-space had a greater impact on the imaging quality than the data at the edges, and then samples the data based on the ideology of normal distribution, i.e., discarding part of the data at the edges of the k-space. The experimental results show that our proposed method can significantly reduce the number of sampled lines and have higher imaging quality with a lower number of ACS lines.

Accelerated three-dimensional susceptibility weighted imaging of the whole spine of fetus at 3T

ObjectivesTo investigate the image quality and capability of generalized auto-calibrating partially parallel acquisition (GRAPPA) accelerated Three-dimensional (3D) susceptibility weighted imaging (SWI) of the whole spine at 3T. MethodsA total of 37 pregnant women (gestation age 22 to 39 weeks, average 29 ± 3 weeks) with suspected fetal vertebral anomalies by ultrasound (US) screening underwent 3.0T MR imaging with 3D SWI, conventional two-dimensional (2D) half-flourier acquisition single-shot turbo spin-echo (HASTE) and 3D true fast imaging with steady-state precession (True FISP). The acquisition time of each protocol was recorded. Signal-to-noise ratios (SNRs) and contrast-to-noise ratios (CNRs) were determined in representative interest regions of fetal thoracic vertebrae and compared among three pulse sequences. Two radiologists rated image quality independently in random order on a 5-point scale. Kappa coefficients were computed to assess inter-observer reliability. Receiver operating characteristic curves were generated, and the area under the curve (AUC) was used to compare the diagnostic performance of each protocol in vertebral deformities. ResultsThe acquisition time was 15 s for 3D-SWI and 17 s for 3D True FISP, significantly shorter than conventional HASTE (37 s; both P < 0.01).Of the three protocols, The SNR was highest on 3D True FISP, while the CNR was highest on 3D SWI. Visualization of all segments of the whole spine by 3D SWI was comparable with 3D True FISP. In contrast, 3D SWI and 3D True FISP depicted cervical and sacrococcygeal vertebrae better than HASTE. The weighted kappa statistic was 0.70–0.89 to evaluate the image quality of all segments of the whole spine, indicating good to excellent interobserver agreement. 3D SWI had the highest diagnostic performance for detecting fetal vertebral anomalies (AUC = 0.92). Conclusions3D-SWI is feasible for improved visualization of the whole fetal vertebral column and its congenital malformations with adequate image quality and high accuracy, thereby providing a supplementary method to conventional MR imaging.

Feasibility study of improved median filtering in PET/MR fusion images with parallel imaging using generalized autocalibrating partially parallel acquisition

This study aimed to analyze the applicability of the improved median filter in positron emission tomography (PET)/magnetic resonance (MR) fusion images based on parallel imaging using generalized autocalibrating partially parallel acquisition (GRAPPA). In this study, a PET/MR fusion imaging system based on a 3.0T magnetic field and 18F radioisotope were used. An improved median filter that can set a mask of the median value more efficiently than before was modeled and applied to the acquired image. As quantitative evaluation parameters of the noise level, the contrast to noise ratio (CNR) and coefficient of variation (COV) were calculated. Additionally, no-reference-based evaluation parameters were used to analyze the overall image quality. We confirmed that the CNR and COV values of the PET/MR fusion images to which the improved median filter was applied improved by approximately 3.32 and 2.19 times on average, respectively, compared to the noisy image. In addition, the no-reference-based evaluation results showed a similar trend for the noise-level results. In conclusion, we demonstrated that it can be supplemented by using an improved median filter, which suggests the problem of image quality degradation of PET/MR fusion images that shortens scan time using GRAPPA.

A miniature U-net for k-space-based parallel magnetic resonance imaging reconstruction with a mixed loss function.

BackgroundDeep learning-based magnetic resonance imaging (MRI) methods require in most cases a separate dataset with thousands of images for each anatomical site to train the network model. This paper proposes a miniature U-net method for k-space-based parallel MRI where the network model is trained individually for each scan using scan-specific autocalibrating signal data.MethodsThe original U-net was tailored with fewer layers and channels, and the network was trained using the autocalibrating signal data with a mixing loss function involving magnitude loss and phase loss. The performance of the proposed method was measured using both phantom and in vivo datasets compared to scan-specific robust artificial-neural-networks for k-space interpolation (RAKI) and generalized autocalibrating partially parallel acquisitions (GRAPPA).ResultsThe proposed method alleviates aliasing artifacts and reduces noise with an acceleration factor of four for phantom and in vivo data. Compared with RAKI and GRAPPA, the proposed method represents an improvement with a structural similarity index measure of between 0.02 and 0.05 and a peak signal-to-noise ratio (PSNR) of between 0.1 and 3.ConclusionsThe proposed method introduces a miniature U-net to reconstruct the missing k-space data, which can provide an optimal trade-off between network performance and requirement of training samples. Experimental results indicate that the proposed method can improve image quality compared with the deep learning-based k-space parallel magnitude resonance imaging method.

Impact of autocalibration method on accelerated EPI of the cervical spinal cord at 7 T.

The spinal cord contains sensorimotor neural circuits of scientific and clinical interest. However, spinal cord functional MRI (fMRI) is significantly more technically demanding than brain fMRI, due primarily to its proximity to the lungs. Accelerated echo-planar imaging (EPI) at 7 T is particularly vulnerable to k-space phase inconsistencies induced by motion or B0 fluctuation, during either autocalibration signal (ACS) or time-series acquisition. For 7 T brain fMRI, sensitivity to motion and B0 fluctuation can be reduced using a re-ordered segmented EPI ACS based on the fast low-angle excitation echo-planar technique (FLEET). However, respiration-induced B0 fluctuations (exceeding 100 Hz at C7) are greater, and fewer k-space lines per slice are required for cervical spinal cord fMRI at 7 T, necessitating a separate evaluation of ACS methods. We compared 24-line single-shot EPI with 48-line two-shot segmented EPI, two-shot FLEET, and gradient echo (GRE)-based ACS acquisition methods, performed under various physiological conditions, in terms of temporal signal-to-noise ratio and prevalence of artifacts in generalized autocalibrating partially parallel acquisition (GRAPPA)-accelerated EPI of the cervical spinal cord at 7 T. Segmented EPI and FLEET ACS produce images with nearly identical patterns of severe image artifacts. GRE and single-shot EPI ACS consistently produce images free from significant artifacts, and temporal signal-to-noise ratio is significantly greater for GRE ACS, particularly in lower slices where through-slice dephasing is most severe. GRE and single-shot EPI-ACS acquisition methods, which are robust to respiration-induced phase errors between k-space segments, produce images with fewer and less severe artifacts than either FLEET or conventionally segmented EPI for accelerated EPI of the cervical spinal cord at7T.

Multi-Domain Neumann Network with Sensitivity Maps for Parallel MRI Reconstruction.

MRI is an imaging technology that non-invasively obtains high-quality medical images for diagnosis. However, MRI has the major disadvantage of long scan times which cause patient discomfort and image artifacts. As one of the methods for reducing the long scan time of MRI, the parallel MRI method for reconstructing a high-fidelity MR image from under-sampled multi-coil k-space data is widely used. In this study, we propose a method to reconstruct a high-fidelity MR image from under-sampled multi-coil k-space data using deep-learning. The proposed multi-domain Neumann network with sensitivity maps (MDNNSM) is based on the Neumann network and uses a forward model including coil sensitivity maps for parallel MRI reconstruction. The MDNNSM consists of three main structures: the CNN-based sensitivity reconstruction block estimates coil sensitivity maps from multi-coil under-sampled k-space data; the recursive MR image reconstruction block reconstructs the MR image; and the skip connection accumulates each output and produces the final result. Experiments using the fastMRI T1-weighted brain image dataset were conducted at acceleration factors of 2, 4, and 8. Qualitative and quantitative experimental results show that the proposed MDNNSM method reconstructs MR images more accurately than other methods, including the generalized autocalibrating partially parallel acquisitions (GRAPPA) method and the original Neumann network.

Clinical Application of MPRAGE Wave Controlled Aliasing in Parallel Imaging (Wave-CAIPI): A Comparative Study with MPRAGE GRAPPA.

To compare reliability and elucidate clinical application of magnetization-prepared rapid gradient-echo (MPRAGE) with 9-fold acceleration by using wave-controlled aliasing in parallel imaging (Wave-CAIPI 3 × 3) in comparison to conventional MPRAGE accelerated by using generalized autocalibrating partially parallel acquisition (GRAPPA) 2 × 1. A total of 26 healthy volunteers and 33 patients were included in this study. Subjects were scanned with two MPRAGEs, GRAPPA 2 × 1 and Wave-CAIPI 3 × 3 acquired in 5 min 21 s and 1 min 42 s, respectively, on a 3T MR scanner. Healthy volunteers underwent additional two MPRAGEs (CAIPI 3 × 3 and GRAPPA 3 × 3). The image quality of the four MPRAGEs was visually evaluated with a 5-point scale in healthy volunteers, and the SNR of four MPRAGEs was also calculated by measuring the phantom 10 times with each MPRAGE. Based on the results of the visual evaluation, voxel-based morphometry (VBM) analyses, including subfield analysis, were performed only for GRAPPA 2 × 1 and Wave-CAIPI 3 × 3. Correlation of segmentation results between GRAPPA 2 × 1 and Wave-CAIPI 3 × 3 was assessed. In visual evaluations, scores for MPRAGE GRAPPA 2 × 1 (mean rank: 4.00) were significantly better than those for Wave-CAIPI 3 × 3 (mean rank: 3.00), CAIPI 3 × 3 (mean rank: 1.83), and GRAPPA 3 × 3 (mean rank: 1.17), and scores for Wave-CAIPI 3×3 were significantly better than those for CAIPI 3 × 3 and GRAPPA 3 × 3. Image noise was evident at the center for additional MPRAGE CAIPI 3 × 3 and GRAPPA 3 × 3. The correlation of segmentation results between GRAPPA 2 × 1 and Wave-CAIPI 3 × 3 was higher than 0.85 in all VOIs except globus pallidus. Subfield analysis of hippocampus also showed a high correlation between GRAPPA 2 × 1 and Wave-CAIPI 3 × 3. MPRAGE Wave-CAIPI 3 × 3 shows relatively better contrast, despite of its short scan time of 1 min 42 s. The volumes derived from automated segmentation of MPRAGE Wave-CAIPI are considered to be reliable measures.

A System for Real-Time, Online Mixed-Reality Visualization of Cardiac Magnetic Resonance Images.

Image-guided cardiovascular interventions are rapidly evolving procedures that necessitate imaging systems capable of rapid data acquisition and low-latency image reconstruction and visualization. Compared to alternative modalities, Magnetic Resonance Imaging (MRI) is attractive for guidance in complex interventional settings thanks to excellent soft tissue contrast and large fields-of-view without exposure to ionizing radiation. However, most clinically deployed MRI sequences and visualization pipelines exhibit poor latency characteristics, and spatial integration of complex anatomy and device orientation can be challenging on conventional 2D displays. This work demonstrates a proof-of-concept system linking real-time cardiac MR image acquisition, online low-latency reconstruction, and a stereoscopic display to support further development in real-time MR-guided intervention. Data are acquired using an undersampled, radial trajectory and reconstructed via parallelized through-time radial generalized autocalibrating partially parallel acquisition (GRAPPA) implemented on graphics processing units. Images are rendered for display in a stereoscopic mixed-reality head-mounted display. The system is successfully tested by imaging standard cardiac views in healthy volunteers. Datasets comprised of one slice (46 ms), two slices (92 ms), and three slices (138 ms) are collected, with the acquisition time of each listed in parentheses. Images are displayed with latencies of 42 ms/frame or less for all three conditions. Volumetric data are acquired at one volume per heartbeat with acquisition times of 467 ms and 588 ms when 8 and 12 partitions are acquired, respectively. Volumes are displayed with a latency of 286 ms or less. The faster-than-acquisition latencies for both planar and volumetric display enable real-time 3D visualization of the heart.

Quad-Contrast Imaging: Simultaneous Acquisition of Four Contrast-Weighted Images (PD-Weighted, T₂-Weighted, PD-FLAIR and T₂-FLAIR Images) With Synthetic T₁-Weighted Image, T₁- and T₂-Maps.

Magnetic resonance imaging (MRI) can provide multiple contrast-weighted images using different pulse sequences and protocols. However, a long acquisition time of the images is a major challenge. To address this limitation, a new pulse sequence referred to as quad-contrast imaging is presented. The quad-contrast sequence enables the simultaneous acquisition of four contrast-weighted images (proton density (PD)-weighted, T2-weighted, PD-fluid attenuated inversion recovery (FLAIR), and T2-FLAIR), and the synthesis of T1-weighted images and T1- and T2-maps in a single scan. The scan time is less than 6 min and is further reduced to 2 min 50 s using a deep learning-based parallel imaging reconstruction. The natively acquired quad contrasts demonstrate high quality images, comparable to those from the conventional scans. The deep learning-based reconstruction successfully reconstructed highly accelerated data (acceleration factor 6), reporting smaller normalized root mean squared errors (NRMSEs) and higher structural similarities (SSIMs) than those from conventional generalized autocalibrating partially parallel acquisitions (GRAPPA)-reconstruction (mean NRMSE of 4.36% vs. 10.54% and mean SSIM of 0.990 vs. 0.953). In particular, the FLAIR contrast is natively acquired and does not suffer from lesion-like artifacts at the boundary of tissue and cerebrospinal fluid, differentiating the proposed method from synthetic imaging methods. The quad-contrast imaging method may have the potentials to be used in a clinical routine as a rapid diagnostic tool.

Scan-specific artifact reduction in k-space (SPARK) neural networks synergize with physics-based reconstruction to accelerate MRI.

To develop a scan-specific model that estimates and corrects k-space errors made when reconstructing accelerated MRI data. Scan-specific artifact reduction in k-space (SPARK) trains a convolutional-neural-network to estimate and correct k-space errors made by an input reconstruction technique by back-propagating from the mean-squared-error loss between an auto-calibration signal (ACS) and the input technique's reconstructed ACS. First, SPARK is applied to generalized autocalibrating partially parallel acquisitions (GRAPPA) and demonstrates improved robustness over other scan-specific models, such as robust artificial-neural-networks for k-space interpolation (RAKI) and residual-RAKI. Subsequent experiments demonstrate that SPARK synergizes with residual-RAKI to improve reconstruction performance. SPARK also improves reconstruction quality when applied to advanced acquisition and reconstruction techniques like 2D virtual coil (VC-) GRAPPA, 2D LORAKS, 3D GRAPPA without an integrated ACS region, and 2D/3D wave-encoded imaging. SPARK yields SSIM improvement and 1.5 - 2× root mean squared error (RMSE) reduction when applied to GRAPPA and improves robustness to ACS size for various acceleration rates in comparison to other scan-specific techniques. When applied to advanced reconstruction techniques such as residual-RAKI, 2D VC-GRAPPA and LORAKS, SPARK achieves up to 20% RMSE improvement. SPARK with 3D GRAPPA also improves RMSE performance by ~2×, SSIM performance, and perceived image quality without a fully sampled ACS region. Finally, SPARK synergizes with non-Cartesian, 2D and 3D wave-encoding imaging by reducing RMSE between 20% and 25% and providing qualitative improvements. SPARK synergizes with physics-based acquisition and reconstruction techniques to improve accelerated MRI by training scan-specific models to estimate and correct reconstruction errors in k-space.

Comparative analysis of the kernel types in generalized autocalibrating partially parallel acquisition algorithms for accelerated three‐dimensional magnetic resonance spectroscopic imaging data

AbstractThis study was conducted to comparatively analyze the performance of generalized autocalibrating partially parallel acquisition (GRAPPA) algorithms according to the kernel types, which has not yet been reported. A GRAPPA algorithm using one 3D kernel or other kernels was used to reconstruct the undersampled 3D MRSI data along three or two PE directions. To quantitatively assess the quality of the reconstruction, the normalized root–mean–squared errors (nRMSEs) were calculated in simulated, phantom, and in vivo data. The nRMSE of the GRAPPA algorithm using one 3D kernel in reconstructing undersampled 3D MRSI data along three PE directions was the lowest. The spectra and maps from the reconstructed data using one 3D kernel were the closest to those from the fully sampled data. We recommend acquiring 3D MRSI data accelerated along three PE directions to reduce scan time and indispensably applying the GRAPPA algorithm using one 3Dkernel to improve reconstruction performance.