High-quality AFM image acquisition of living cells by modified residual encoder-decoder network.

Atomic force microscope enables ultra-precision imaging of living cells. However, atomic force microscope imaging is a complex and time-consuming process. The obtained images of living cells usually have low resolution and are easily influenced by noise leading to unsatisfactory imaging quality, obstructing the research and analysis based on cell images. Herein, an adaptive attention image reconstruction network based on residual encoder-decoder was proposed, through the combination of deep learning technology and atomic force microscope imaging supporting high-quality cell image acquisition. Compared with other learning-based methods, the proposed network showed higher peak signal-to-noise ratio, higher structural similarity and better image reconstruction performances. In addition, the cell images reconstructed by each method were used for cell recognition, and the cell images reconstructed by the proposed network had the highest cell recognition rate. The proposed network has brought insights into the atomic force microscope-based imaging of living cells and cell image reconstruction, which is of great significance in biological and medical research.

DNA-QLC: an efficient and reliable image encoding scheme for DNA storage.

BACKGROUND: DNA storage has the advantages of large capacity, long-term stability, and low power consumption relative to other storage mediums, making it a promising new storage medium for multimedia information such as images. However, DNA storage has a low coding density and weak error correction ability. RESULTS: To achieve more efficient DNA storage image reconstruction, we propose DNA-QLC (QRes-VAE and Levenshtein code (LC)), which uses the quantized ResNet VAE (QRes-VAE) model and LC for image compression and DNA sequence error correction, thus improving both the coding density and error correction ability. Experimental results show that the DNA-QLC encoding method can not only obtain DNA sequences that meet the combinatorial constraints, but also have a net information density that is 2.4 times higher than DNA Fountain. Furthermore, at a higher error rate (2%), DNA-QLC achieved image reconstruction with an SSIM value of 0.917. CONCLUSIONS: The results indicate that the DNA-QLC encoding scheme guarantees the efficiency and reliability of the DNA storage system and improves the application potential of DNA storage for multimedia information such as images.

The "hidden noise" problem in MR image reconstruction.

PURPOSE: The performance of modern image reconstruction methods is commonly judged using quantitative error metrics like root mean squared-error and the structural similarity index, which are calculated by comparing reconstructed images against fully sampled reference data. In practice, the reference data will contain noise and is not a true gold standard. In this work, we demonstrate that the "hidden noise" present in reference data can substantially confound standard approaches for ranking different image reconstruction results. METHODS: Using both experimental and simulated k-space data and several different image reconstruction techniques, we examined whether there was correlation between performance metrics obtained with typical noisy reference data versus those obtained with higher-quality reference data. RESULTS: For conventional performance metrics, the reconstructions that matched best with the higher-quality reference data were substantially different from the reconstructions that matched best with typical noisy reference data. This leads to suboptimal reconstruction results if the performance with respect to noisy reference data is used to select which reconstruction methods/parameters to employ. These issues were reduced when employing alternative error metrics that better account for noise. CONCLUSION: Reference data containing hidden noise can substantially mislead the ranking of image reconstruction methods when using conventional error metrics, but this issue can be mitigated with alternative error metrics.

Robust multishot diffusion-weighted imaging of the abdomen with region-based shot rejection.

PURPOSE: Diffusion-weighted (DW) imaging provides a useful clinical contrast, but is susceptible to motion-induced dephasing caused by the application of strong diffusion gradients. Phase navigators are commonly used to resolve shot-to-shot motion-induced phase in multishot reconstructions, but poor phase estimates result in signal dropout and Apparent Diffusion Coefficient (ADC) overestimation. These artifacts are prominent in the abdomen, a region prone to involuntary cardiac and respiratory motion. To improve the robustness of DW imaging in the abdomen, region-based shot rejection schemes that selectively weight regions where the shot-to-shot phase is poorly estimated were evaluated. METHODS: Spatially varying weights for each shot, reflecting both the accuracy of the estimated phase and the degree of subvoxel dephasing, were estimated from the phase navigator magnitude images. The weighting was integrated into a multishot reconstruction using different formulations and phase navigator resolutions and tested with different phase navigator resolutions in multishot DW-echo Planar Imaging acquisitions of the liver and pancreas, using conventional monopolar and velocity-compensated diffusion encoding. Reconstructed images and ADC estimates were compared qualitatively. RESULTS: The proposed region-based shot rejection reduces banding and signal dropout artifacts caused by physiological motion in the liver and pancreas. Shot rejection allows conventional monopolar diffusion encoding to achieve median ADCs in the pancreas comparable to motion-compensated diffusion encoding, albeit with a greater spread of ADCs. CONCLUSION: Region-based shot rejection is a linear reconstruction that improves the motion robustness of multi-shot DWI and requires no sequence modifications.

HyperSLICE: HyperBand optimized spiral for low-latency interactive cardiac examination.

PURPOSE: Interactive cardiac MRI is used for fast scan planning and MR-guided interventions. However, the requirement for real-time acquisition and near-real-time visualization constrains the achievable spatio-temporal resolution. This study aims to improve interactive imaging resolution through optimization of undersampled spiral sampling and leveraging of deep learning for low-latency reconstruction (deep artifact suppression). METHODS: A variable density spiral trajectory was parametrized and optimized via HyperBand to provide the best candidate trajectory for rapid deep artifact suppression. Training data consisted of 692 breath-held CINEs. The developed interactive sequence was tested in simulations and prospectively in 13 subjects (10 for image evaluation, 2 during catheterization, 1 during exercise). In the prospective study, the optimized framework-HyperSLICE- was compared with conventional Cartesian real-time and breath-hold CINE imaging in terms quantitative and qualitative image metrics. Statistical differences were tested using Friedman chi-squared tests with post hoc Nemenyi test (p < 0.05). RESULTS: In simulations the normalized RMS error, peak SNR, structural similarity, and Laplacian energy were all statistically significantly higher using optimized spiral compared to radial and uniform spiral sampling, particularly after scan plan changes (structural similarity: 0.71 vs. 0.45 and 0.43). Prospectively, HyperSLICE enabled a higher spatial and temporal resolution than conventional Cartesian real-time imaging. The pipeline was demonstrated in patients during catheter pull back, showing sufficiently fast reconstruction for interactive imaging. CONCLUSION: HyperSLICE enables high spatial and temporal resolution interactive imaging. Optimizing the spiral sampling enabled better overall image quality and superior handling of image transitions compared with radial and uniform spiral trajectories.

Effect of MR head coil geometry on deep-learning-based MR image reconstruction.

PURPOSE: To investigate whether parallel imaging-imposed geometric coil constraints can be relaxed when using a deep learning (DL)-based image reconstruction method as opposed to a traditional non-DL method. THEORY AND METHODS: Traditional and DL-based MR image reconstruction approaches operate in fundamentally different ways: Traditional methods solve a system of equations derived from the image data whereas DL methods use data/target pairs to learn a generalizable reconstruction model. Two sets of head coil profiles were evaluated: (1) 8-channel and (2) 32-channel geometries. A DL model was compared to conjugate gradient SENSE (CG-SENSE) and L1-wavelet compressed sensing (CS) through quantitative metrics and visual assessment as coil overlap was increased. RESULTS: Results were generally consistent between experiments. As coil overlap increased, there was a significant (p < 0.001) decrease in performance in most cases for all methods. The decrease was most pronounced for CG-SENSE, and the DL models significantly outperformed (p < 0.001) their non-DL counterparts in all scenarios. CS showed improved robustness to coil overlap and signal-to-noise ratio (SNR) versus CG-SENSE, but had quantitatively and visually poorer reconstructions characterized by blurriness as compared to DL. DL showed virtually no change in performance across SNR and very small changes across coil overlap. CONCLUSION: The DL image reconstruction method produced images that were robust to coil overlap and of higher quality than CG-SENSE and CS. This suggests that geometric coil design constraints can be relaxed when using DL reconstruction methods.

Improving Xenon-129 lung ventilation image SNR with deep-learning based image reconstruction.

PURPOSE: To evaluate the feasibility and utility of a deep learning (DL)-based reconstruction for improving the SNR of hyperpolarized 129Xe lung ventilation MRI. METHODS: 129Xe lung ventilation MRI data acquired from patients with asthma and/or chronic obstructive pulmonary disease (COPD) were retrospectively reconstructed with a commercial DL reconstruction pipeline at five different denoising levels. Quantitative imaging metrics of lung ventilation including ventilation defect percentage (VDP) and ventilation heterogeneity index (VHI) were compared between each set of DL-reconstructed images and alternative denoising strategies including: filtering, total variation denoising and higher-order singular value decomposition. Structural similarity between the denoised and original images was assessed. In a prospective study, the feasibility of using SNR gains from DL reconstruction to allow natural-abundance xenon MRI was evaluated in healthy volunteers. RESULTS: 129Xe ventilation image SNR was improved with DL reconstruction when compared with conventionally reconstructed images. In patients with asthma and/or COPD, DL-reconstructed images exhibited a slight positive bias in ventilation defect percentage (1.3% at 75% denoising) and ventilation heterogeneity index (˜1.4) when compared with conventionally reconstructed images. Additionally, DL-reconstructed images preserved structural similarity more effectively than data denoised using alternative approaches. DL reconstruction greatly improved image SNR (greater than threefold), to a level that 129Xe ventilation imaging using natural-abundance xenon appears feasible. CONCLUSION: DL-based image reconstruction significantly improves 129Xe ventilation image SNR, preserves structural similarity, and leads to a minor bias in ventilation metrics that can be attributed to differences in the image sharpness. This tool should help facilitate cost-effective 129Xe ventilation imaging with natural-abundance xenon in the future.

Predictive uncertainty in deep learning-based MR image reconstruction using deep ensembles: Evaluation on the fastMRI data set.

PURPOSE: To estimate pixel-wise predictive uncertainty for deep learning-based MR image reconstruction and to examine the impact of domain shifts and architecture robustness. METHODS: Uncertainty prediction could provide a measure for robustness of deep learning (DL)-based MR image reconstruction from undersampled data. DL methods bear the risk of inducing reconstruction errors like in-painting of unrealistic structures or missing pathologies. These errors may be obscured by visual realism of DL reconstruction and thus remain undiscovered. Furthermore, most methods are task-agnostic and not well calibrated to domain shifts. We propose a strategy that estimates aleatoric (data) and epistemic (model) uncertainty, which entails training a deep ensemble (epistemic) with nonnegative log-likelihood (aleatoric) loss in addition to the conventional applied losses terms. The proposed procedure can be paired with any DL reconstruction, enabling investigations of their predictive uncertainties on a pixel level. Five different architectures were investigated on the fastMRI database. The impact on the examined uncertainty of in-distributional and out-of-distributional data with changes to undersampling pattern, imaging contrast, imaging orientation, anatomy, and pathology were explored. RESULTS: Predictive uncertainty could be captured and showed good correlation to normalized mean squared error. Uncertainty was primarily focused along the aliased anatomies and on hyperintense and hypointense regions. The proposed uncertainty measure was able to detect disease prevalence shifts. Distinct predictive uncertainty patterns were observed for changing network architectures. CONCLUSION: The proposed approach enables aleatoric and epistemic uncertainty prediction for DL-based MR reconstruction with an interpretable examination on a pixel level.

SPICER: Self-supervised learning for MRI with automatic coil sensitivity estimation and reconstruction.

PURPOSE: To introduce a novel deep model-based architecture (DMBA), SPICER, that uses pairs of noisy and undersampled k-space measurements of the same object to jointly train a model for MRI reconstruction and automatic coil sensitivity estimation. METHODS: SPICER consists of two modules to simultaneously reconstructs accurate MR images and estimates high-quality coil sensitivity maps (CSMs). The first module, CSM estimation module, uses a convolutional neural network (CNN) to estimate CSMs from the raw measurements. The second module, DMBA-based MRI reconstruction module, forms reconstructed images from the input measurements and the estimated CSMs using both the physical measurement model and learned CNN prior. With the benefit of our self-supervised learning strategy, SPICER can be efficiently trained without any fully sampled reference data. RESULTS: We validate SPICER on both open-access datasets and experimentally collected data, showing that it can achieve state-of-the-art performance in highly accelerated data acquisition settings (up to 10 × $$ 10\times $$ ). Our results also highlight the importance of different modules of SPICER-including the DMBA, the CSM estimation, and the SPICER training loss-on the final performance of the method. Moreover, SPICER can estimate better CSMs than pre-estimation methods especially when the ACS data is limited. CONCLUSION: Despite being trained on noisy undersampled data, SPICER can reconstruct high-quality images and CSMs in highly undersampled settings, which outperforms other self-supervised learning methods and matches the performance of the well-known E2E-VarNet trained on fully sampled ground-truth data.

Self-supervised learning for improved calibrationless radial MRI with NLINV-Net.

PURPOSE: To develop a neural network architecture for improved calibrationless reconstruction of radial data when no ground truth is available for training. METHODS: NLINV-Net is a model-based neural network architecture that directly estimates images and coil sensitivities from (radial) k-space data via nonlinear inversion (NLINV). Combined with a training strategy using self-supervision via data undersampling (SSDU), it can be used for imaging problems where no ground truth reconstructions are available. We validated the method for (1) real-time cardiac imaging and (2) single-shot subspace-based quantitative T1 mapping. Furthermore, region-optimized virtual (ROVir) coils were used to suppress artifacts stemming from outside the field of view and to focus the k-space-based SSDU loss on the region of interest. NLINV-Net-based reconstructions were compared with conventional NLINV and PI-CS (parallel imaging + compressed sensing) reconstruction and the effect of the region-optimized virtual coils and the type of training loss was evaluated qualitatively. RESULTS: NLINV-Net-based reconstructions contain significantly less noise than the NLINV-based counterpart. ROVir coils effectively suppress streakings which are not suppressed by the neural networks while the ROVir-based focused loss leads to visually sharper time series for the movement of the myocardial wall in cardiac real-time imaging. For quantitative imaging, T1-maps reconstructed using NLINV-Net show similar quality as PI-CS reconstructions, but NLINV-Net does not require slice-specific tuning of the regularization parameter. CONCLUSION: NLINV-Net is a versatile tool for calibrationless imaging which can be used in challenging imaging scenarios where a ground truth is not available.

Self-calibrated subspace reconstruction for multidimensional MR fingerprinting for simultaneous relaxation and diffusion quantification.

PURPOSE: To propose a new reconstruction method for multidimensional MR fingerprinting (mdMRF) to address shading artifacts caused by physiological motion-induced measurement errors without navigating or gating. METHODS: The proposed method comprises two procedures: self-calibration and subspace reconstruction. The first procedure (self-calibration) applies temporally local matrix completion to reconstruct low-resolution images from a subset of under-sampled data extracted from the k-space center. The second procedure (subspace reconstruction) utilizes temporally global subspace reconstruction with pre-estimated temporal subspace from low-resolution images to reconstruct aliasing-free, high-resolution, and time-resolved images. After reconstruction, a customized outlier detection algorithm was employed to automatically detect and remove images corrupted by measurement errors. Feasibility, robustness, and scan efficiency were evaluated through in vivo human brain imaging experiments. RESULTS: The proposed method successfully reconstructed aliasing-free, high-resolution, and time-resolved images, where the measurement errors were accurately represented. The corrupted images were automatically and robustly detected and removed. Artifact-free T1, T2, and ADC maps were generated simultaneously. The proposed reconstruction method demonstrated robustness across different scanners, parameter settings, and subjects. A high scan efficiency of less than 20 s per slice has been achieved. CONCLUSION: The proposed reconstruction method can effectively alleviate shading artifacts caused by physiological motion-induced measurement errors. It enables simultaneous and artifact-free quantification of T1, T2, and ADC using mdMRF scans without prospective gating, with robustness and high scan efficiency.

A decay-modeled compressed sensing reconstruction approach for non-Cartesian hyperpolarized 129Xe MRI.

PURPOSE: Hyperpolarized 129Xe MRI benefits from non-Cartesian acquisitions that sample k-space efficiently and rapidly. However, their reconstructions are complex and burdened by decay processes unique to hyperpolarized gas. Currently used gridded reconstructions are prone to artifacts caused by magnetization decay and are ill-suited for undersampling. We present a compressed sensing (CS) reconstruction approach that incorporates magnetization decay in the forward model, thereby producing images with increased sharpness and contrast, even in undersampled data. METHODS: Radio-frequency, T1, and T 2 * $$ {\mathrm{T}}_2^{\ast } $$ decay processes were incorporated into the forward model and solved using iterative methods including CS. The decay-modeled reconstruction was validated in simulations and then tested in 2D/3D-spiral ventilation and 3D-radial gas-exchange MRI. Quantitative metrics including apparent-SNR and sharpness were compared between gridded, CS, and twofold undersampled CS reconstructions. Observations were validated in gas-exchange data collected from 15 healthy and 25 post-hematopoietic-stem-cell-transplant participants. RESULTS: CS reconstructions in simulations yielded images with threefold increases in accuracy. CS increased sharpness and contrast for ventilation in vivo imaging and showed greater accuracy for undersampled acquisitions. CS improved gas-exchange imaging, particularly in the dissolved-phase where apparent-SNR improved, and structure was made discernable. Finally, CS showed repeatability in important global gas-exchange metrics including median dissolved-gas signal ratio and median angle between real/imaginary components. CONCLUSION: A non-Cartesian CS reconstruction approach that incorporates hyperpolarized 129Xe decay processes is presented. This approach enables improved image sharpness, contrast, and overall image quality in addition to up-to threefold undersampling. This contribution benefits all hyperpolarized gas MRI through improved accuracy and decreased scan durations.

Sampling pattern discrepancy in the application of compressed sensing hyperpolarized xenon-129 lung MRI.

Although hyperpolarized (HP) 129Xe ventilation MRI can be carried out within a breath hold, it is still challenging for many sick patients. Compressed sensing (CS) is a viable alternative to accelerate this approach. However, undersampled images with identical sampling ratios differ from one another. Twenty subjects (n = 10 healthy and n = 10 patients with asthma) were scanned using a GE MR750 3 T scanner, acquiring fully sampled 2D multi-slice HP 129Xe lung ventilation images (10 s breath hold, 128 × 80 (FE × PE-frequency encoding × phase encoding) and 16 slices). Using fully sampled data, 500 variable-density Cartesian random undersampling patterns were generated, each at eight different sampling ratios from 10% to 80%. The parallel imaging and compressed sensing (PICS) command from BART was employed to reconstruct undersampled data. The signal to noise ratio (SNR), structural similarity index measurement (SSIM) and sidelobe to peak ratio of each were subsequently compared. There was a high degree of variation in both SNR and SSIM results from each of the 500 masks of each sampling rate. As the undersampling increases, there is more variation in the quantifying metrics, for both healthy and asthmatic individuals. Our study shows that random undersampling poses a significant challenge when applied at sampling ratios less than 60%, despite fulfilling CS's incoherency criteria. Such low sampling ratios will result in a large variety of undersampling patterns. Therefore, skipped segments of k-space cannot be allowed to happen randomly at low sampling rates. By optimizing the sampling pattern, CS will reach its full potential and be able to be applied to a highly undersampled 129Xe lung dataset.

Ferumoxytol-Enhanced Cardiac Cine MRI Reconstruction Using a Variable-Splitting Spatiotemporal Network.

BACKGROUND: Balanced steady-state free precession (bSSFP) imaging is commonly used in cardiac cine MRI but prone to image artifacts. Ferumoxytol-enhanced (FE) gradient echo (GRE) has been proposed as an alternative. Utilizing the abundance of bSSFP images to develop a computationally efficient network that is applicable to FE GRE cine would benefit future network development. PURPOSE: To develop a variable-splitting spatiotemporal network (VSNet) for image reconstruction, trained on bSSFP cine images and applicable to FE GRE cine images. STUDY TYPE: Retrospective and prospective. SUBJECTS: 41 patients (26 female, 53 ± 19 y/o) for network training, 31 patients (19 female, 49 ± 17 y/o) and 5 healthy subjects (5 female, 30 ± 7 y/o) for testing. FIELD STRENGTH/SEQUENCE: 1.5T and 3T, bSSFP and GRE. ASSESSMENT: VSNet was compared to VSNet with total variation loss, compressed sensing and low rank methods for 14× accelerated data. The GRAPPA×2/×3 images served as the reference. Peak signal-to-noise-ratio (PSNR), structural similarity index (SSIM), left ventricular (LV) and right ventricular (RV) end-diastolic volume (EDV), end-systolic volume (ESV), and ejection fraction (EF) were measured. Qualitative image ranking and scoring were independently performed by three readers. Latent scores were calculated based on scores of each method relative to the reference. STATISTICS: Linear mixed-effects regression, Tukey method, Fleiss' Kappa, Bland-Altman analysis, and Bayesian categorical cumulative probit model. A P-value <0.05 was considered statistically significant. RESULTS: VSNet achieved significantly higher PSNR (32.7 ± 0.2), SSIM (0.880 ± 0.004), rank (2.14 ± 0.06), and latent scores (-1.72 ± 0.22) compared to other methods (rank >2.90, latent score < -2.63). Fleiss' Kappa was 0.52 for scoring and 0.61 for ranking. VSNet showed no significantly different LV and RV ESV (P = 0.938) and EF (P = 0.143) measurements, but statistically significant different (2.62 mL) EDV measurements compared to the reference. CONCLUSION: VSNet produced the highest image quality and the most accurate functional measurements for FE GRE cine images among the tested 14× accelerated reconstruction methods. LEVEL OF EVIDENCE: 3 TECHNICAL EFFICACY: Stage 1.

Surpassing light inhomogeneities in structured-illumination microscopy with FlexSIM.

Super-resolution structured-illumination microscopy (SIM) is a powerful technique that allows one to surpass the diffraction limit by up to a factor two. Yet, its practical use is hampered by its sensitivity to imaging conditions which makes it prone to reconstruction artefacts. In this work, we present FlexSIM, a flexible SIM reconstruction method capable to handle highly challenging data. Specifically, we demonstrate the ability of FlexSIM to deal with the distortion of patterns, the high level of noise encountered in live imaging, as well as out-of-focus fluorescence. Moreover, we show that FlexSIM achieves state-of-the-art performance over a variety of open SIM datasets.

A motion-corrected deep-learning reconstruction framework for accelerating whole-heart magnetic resonance imaging in patients with congenital heart disease.

BACKGROUND: Cardiovascular magnetic resonance (CMR) is an important imaging modality for the assessment and management of adult patients with congenital heart disease (CHD). However, conventional techniques for three-dimensional (3D) whole-heart acquisition involve long and unpredictable scan times and methods that accelerate scans via k-space undersampling often rely on long iterative reconstructions. Deep-learning-based reconstruction methods have recently attracted much interest due to their capacity to provide fast reconstructions while often outperforming existing state-of-the-art methods. In this study, we sought to adapt and validate a non-rigid motion-corrected model-based deep learning (MoCo-MoDL) reconstruction framework for 3D whole-heart MRI in a CHD patient cohort. METHODS: The previously proposed deep-learning reconstruction framework MoCo-MoDL, which incorporates a non-rigid motion-estimation network and a denoising regularization network within an unrolled iterative reconstruction, was trained in an end-to-end manner using 39 CHD patient datasets. Once trained, the framework was evaluated in eight CHD patient datasets acquired with seven-fold prospective undersampling. Reconstruction quality was compared with the state-of-the-art non-rigid motion-corrected patch-based low-rank reconstruction method (NR-PROST) and against reference images (acquired with three-or-four-fold undersampling and reconstructed with NR-PROST). RESULTS: Seven-fold undersampled scan times were 2.1 ± 0.3 minutes and reconstruction times were ∼30 seconds, approximately 240 times faster than an NR-PROST reconstruction. Image quality comparable to the reference images was achieved using the proposed MoCo-MoDL framework, with no statistically significant differences found in any of the assessed quantitative or qualitative image quality measures. Additionally, expert image quality scores indicated the MoCo-MoDL reconstructions were consistently of a higher quality than the NR-PROST reconstructions of the same data, with the differences in 12 of the 22 scores measured for individual vascular structures found to be statistically significant. CONCLUSION: The MoCo-MoDL framework was applied to an adult CHD patient cohort, achieving good quality 3D whole-heart images from ∼2-minute scans with reconstruction times of ∼30 seconds.

Deep learning for accelerated and robust MRI reconstruction.

Deep learning (DL) has recently emerged as a pivotal technology for enhancing magnetic resonance imaging (MRI), a critical tool in diagnostic radiology. This review paper provides a comprehensive overview of recent advances in DL for MRI reconstruction, and focuses on various DL approaches and architectures designed to improve image quality, accelerate scans, and address data-related challenges. It explores end-to-end neural networks, pre-trained and generative models, and self-supervised methods, and highlights their contributions to overcoming traditional MRI limitations. It also discusses the role of DL in optimizing acquisition protocols, enhancing robustness against distribution shifts, and tackling biases. Drawing on the extensive literature and practical insights, it outlines current successes, limitations, and future directions for leveraging DL in MRI reconstruction, while emphasizing the potential of DL to significantly impact clinical imaging practices.

Large-scale 3D non-Cartesian coronary MRI reconstruction using distributed memory-efficient physics-guided deep learning with limited training data.

OBJECT: To enable high-quality physics-guided deep learning (PG-DL) reconstruction of large-scale 3D non-Cartesian coronary MRI by overcoming challenges of hardware limitations and limited training data availability. MATERIALS AND METHODS: While PG-DL has emerged as a powerful image reconstruction method, its application to large-scale 3D non-Cartesian MRI is hindered by hardware limitations and limited availability of training data. We combine several recent advances in deep learning and MRI reconstruction to tackle the former challenge, and we further propose a 2.5D reconstruction using 2D convolutional neural networks, which treat 3D volumes as batches of 2D images to train the network with a limited amount of training data. Both 3D and 2.5D variants of the PG-DL networks were compared to conventional methods for high-resolution 3D kooshball coronary MRI. RESULTS: Proposed PG-DL reconstructions of 3D non-Cartesian coronary MRI with 3D and 2.5D processing outperformed all conventional methods both quantitatively and qualitatively in terms of image assessment by an experienced cardiologist. The 2.5D variant further improved vessel sharpness compared to 3D processing, and scored higher in terms of qualitative image quality. DISCUSSION: PG-DL reconstruction of large-scale 3D non-Cartesian MRI without compromising image size or network complexity is achieved, and the proposed 2.5D processing enables high-quality reconstruction with limited training data.

Deep learning image reconstruction generates thinner slice iodine maps with improved image quality to increase diagnostic acceptance and lesion conspicuity: a prospective study on abdominal dual-energy CT.

BACKGROUND: To assess the improvement of image quality and diagnostic acceptance of thinner slice iodine maps enabled by deep learning image reconstruction (DLIR) in abdominal dual-energy CT (DECT). METHODS: This study prospectively included 104 participants with 136 lesions. Four series of iodine maps were generated based on portal-venous scans of contrast-enhanced abdominal DECT: 5-mm and 1.25-mm using adaptive statistical iterative reconstruction-V (Asir-V) with 50% blending (AV-50), and 1.25-mm using DLIR with medium (DLIR-M), and high strength (DLIR-H). The iodine concentrations (IC) and their standard deviations of nine anatomical sites were measured, and the corresponding coefficient of variations (CV) were calculated. Noise-power-spectrum (NPS) and edge-rise-slope (ERS) were measured. Five radiologists rated image quality in terms of image noise, contrast, sharpness, texture, and small structure visibility, and evaluated overall diagnostic acceptability of images and lesion conspicuity. RESULTS: The four reconstructions maintained the IC values unchanged in nine anatomical sites (all p > 0.999). Compared to 1.25-mm AV-50, 1.25-mm DLIR-M and DLIR-H significantly reduced CV values (all p < 0.001) and presented lower noise and noise peak (both p < 0.001). Compared to 5-mm AV-50, 1.25-mm images had higher ERS (all p < 0.001). The difference of the peak and average spatial frequency among the four reconstructions was relatively small but statistically significant (both p < 0.001). The 1.25-mm DLIR-M images were rated higher than the 5-mm and 1.25-mm AV-50 images for diagnostic acceptability and lesion conspicuity (all P < 0.001). CONCLUSIONS: DLIR may facilitate the thinner slice thickness iodine maps in abdominal DECT for improvement of image quality, diagnostic acceptability, and lesion conspicuity.

MRI reconstruction with enhanced self-similarity using graph convolutional network.

BACKGROUND: Recent Convolutional Neural Networks (CNNs) perform low-error reconstruction in fast Magnetic Resonance Imaging (MRI). Most of them convolve the image with kernels and successfully explore the local information. Nonetheless, the non-local image information, which is embedded among image patches relatively far from each other, may be lost due to the limitation of the receptive field of the convolution kernel. We aim to incorporate a graph to represent non-local information and improve the reconstructed images by using the Graph Convolutional Enhanced Self-Similarity (GCESS) network. METHODS: First, the image is reconstructed into the graph to extract the non-local self-similarity in the image. Second, GCESS uses spatial convolution and graph convolution to process the information in the image, so that local and non-local information can be effectively utilized. The network strengthens the non-local similarity between similar image patches while reconstructing images, making the reconstruction of structure more reliable. RESULTS: Experimental results on in vivo knee and brain data demonstrate that the proposed method achieves better artifact suppression and detail preservation than state-of-the-art methods, both visually and quantitatively. Under 1D Cartesian sampling with 4 × acceleration (AF = 4), the PSNR of knee data reached 34.19 dB, 1.05 dB higher than that of the compared methods; the SSIM achieved 0.8994, 2% higher than the compared methods. Similar results were obtained for the reconstructed images under other sampling templates as demonstrated in our experiment. CONCLUSIONS: The proposed method successfully constructs a hybrid graph convolution and spatial convolution network to reconstruct images. This method, through its training process, amplifies the non-local self-similarities, significantly benefiting the structural integrity of the reconstructed images. Experiments demonstrate that the proposed method outperforms the state-of-the-art reconstruction method in suppressing artifacts, as well as in preserving image details.