SDnDTI: Self-supervised deep learning-based denoising for diffusion tensor MRI

Improved Diffusion-Weighted Hyperpolarized 129Xe Lung MRI with Patch-Based Higher-Order, Singular Value Decomposition Denoising

RATIONALE AND OBJECTIVES

Hyperpolarized xenon (129Xe) MRI is a noninvasive method to assess pulmonary structure and function. To measure lung microstructure, diffusion-weighted imaging-commonly the apparent diffusion coefficient (ADC)-can be employed to map changes in alveolar-airspace size resulting from normal aging and pulmonary disease. However, low signal-to-noise ratio (SNR) decreases ADC measurement certainty, and biases ADC to spuriously low values. Further, these challenges are most severe in regions of the lung where alveolar simplification or emphysematous remodeling generate abnormally high ADCs. Here, we apply Global Local Higher Order Singular Value Decomposition (GLHOSVD) denoising to enhance image SNR, thereby reducing uncertainty and bias in diffusion measurements.

MATERIALS AND METHODS

GLHOSVD denoising was employed in simulated images and gas phantoms with known diffusion coefficients to validate its effectiveness and optimize parameters for analysis of diffusion-weighted 129Xe MRI. GLHOSVD was applied to data from 120 subjects (34 control, 39 cystic fibrosis (CF), 27 lymphangioleiomyomatosis (LAM), and 20 asthma). Image SNR, ADC, and distributed diffusivity coefficient (DDC) were compared before and after denoising using Wilcoxon signed-rank analysis for all images.

RESULTS

Denoising significantly increased SNR in simulated, phantom, and in-vivo images, showing a greater than 2-fold increase (p < 0.001) across diffusion-weighted images. Although mean ADC and DDC remained unchanged (p > 0.05), ADC and DDC standard deviation decreased significantly in denoised images (p < 0.001).

CONCLUSION

When applied to diffusion-weighted 129Xe images, GLHOSVD improved image quality and allowed airspace size to be quantified in high-diffusion regions of the lungs that were previously inaccessible to measurement due to prohibitively low SNR, thus providing insights into disease pathology.

Field-of-view extension for brain diffusion MRI via deep generative models

Purpose

In brain diffusion magnetic resonance imaging (dMRI), the volumetric and bundle analyses of whole-brain tissue microstructure and connectivity can be severely impeded by an incomplete field of view (FOV). We aim to develop a method for imputing the missing slices directly from existing dMRI scans with an incomplete FOV. We hypothesize that the imputed image with a complete FOV can improve whole-brain tractography for corrupted data with an incomplete FOV. Therefore, our approach provides a desirable alternative to discarding the valuable brain dMRI data, enabling subsequent tractography analyses that would otherwise be challenging or unattainable with corrupted data.

Approach

We propose a framework based on a deep generative model that estimates the absent brain regions in dMRI scans with an incomplete FOV. The model is capable of learning both the diffusion characteristics in diffusion-weighted images (DWIs) and the anatomical features evident in the corresponding structural images for efficiently imputing missing slices of DWIs in the incomplete part of the FOV.

Results

For evaluating the imputed slices, on the Wisconsin Registry for Alzheimer's Prevention (WRAP) dataset, the proposed framework achievedPSNR b 0 = 22.397 ,SSIM b 0 = 0.905 ,PSNR b 1300 = 22.479 , andSSIM b 1300 = 0.893 ; on the National Alzheimer's Coordinating Center (NACC) dataset, it achievedPSNR b 0 = 21.304 ,SSIM b 0 = 0.892 ,PSNR b 1300 = 21.599 , andSSIM b 1300 = 0.877 . The proposed framework improved the tractography accuracy, as demonstrated by an increased average Dice score for 72 tracts ( p < 0.001 ) on both the WRAP and NACC datasets.

Conclusions

Results suggest that the proposed framework achieved sufficient imputation performance in brain dMRI data with an incomplete FOV for improving whole-brain tractography, thereby repairing the corrupted data. Our approach achieved more accurate whole-brain tractography results with an extended and complete FOV and reduced the uncertainty when analyzing bundles associated with Alzheimer's disease.

Zero-DeepSub: Zero-shot deep subspace reconstruction for rapid multiparametric quantitative MRI using 3D-QALAS

PURPOSE

To develop and evaluate methods for (1) reconstructing 3D-quantification using an interleaved Look-Locker acquisition sequence with T2 preparation pulse (3D-QALAS) time-series images using a low-rank subspace method, which enables accurate and rapid T1 and T2 mapping, and (2) improving the fidelity of subspace QALAS by combining scan-specific deep-learning-based reconstruction and subspace modeling.

THEORY AND METHODS

A low-rank subspace method for 3D-QALAS (i.e., subspace QALAS) and zero-shot deep-learning subspace method (i.e., Zero-DeepSub) were proposed for rapid and high fidelity T1 and T2 mapping and time-resolved imaging using 3D-QALAS. Using an ISMRM/NIST system phantom, the accuracy and reproducibility of the T1 and T2 maps estimated using the proposed methods were evaluated by comparing them with reference techniques. The reconstruction performance of the proposed subspace QALAS using Zero-DeepSub was evaluated in vivo and compared with conventional QALAS at high reduction factors of up to nine-fold.

RESULTS

Phantom experiments showed that subspace QALAS had good linearity with respect to the reference methods while reducing biases and improving precision compared to conventional QALAS, especially for T2 maps. Moreover, in vivo results demonstrated that subspace QALAS had better g-factor maps and could reduce voxel blurring, noise, and artifacts compared to conventional QALAS and showed robust performance at up to nine-fold acceleration with Zero-DeepSub, which enabled whole-brain T1, T2, and PD mapping at 1 mm isotropic resolution within 2 min of scan time.

CONCLUSION

The proposed subspace QALAS along with Zero-DeepSub enabled high fidelity and rapid whole-brain multiparametric quantification and time-resolved imaging.

FlexDTI: flexible diffusion gradient encoding scheme-based highly efficient diffusion tensor imaging using deep learning

Objective. Most deep neural network-based diffusion tensor imaging methods require the diffusion gradients' number and directions in the data to be reconstructed to match those in the training data. This work aims to develop and evaluate a novel dynamic-convolution-based method called FlexDTI for highly efficient diffusion tensor reconstruction with flexible diffusion encoding gradient scheme.Approach. FlexDTI was developed to achieve high-quality DTI parametric mapping with flexible number and directions of diffusion encoding gradients. The method used dynamic convolution kernels to embed diffusion gradient direction information into feature maps of the corresponding diffusion signal. Furthermore, it realized the generalization of a flexible number of diffusion gradient directions by setting the maximum number of input channels of the network. The network was trained and tested using datasets from the Human Connectome Project and local hospitals. Results from FlexDTI and other advanced tensor parameter estimation methods were compared.Main results. Compared to other methods, FlexDTI successfully achieves high-quality diffusion tensor-derived parameters even if the number and directions of diffusion encoding gradients change. It reduces normalized root mean squared error by about 50% on fractional anisotropy and 15% on mean diffusivity, compared with the state-of-the-art deep learning method with flexible diffusion encoding gradient scheme.Significance. FlexDTI can well learn diffusion gradient direction information to achieve generalized DTI reconstruction with flexible diffusion gradient scheme. Both flexibility and reconstruction quality can be taken into account in this network.

Enhancing clinical diagnostics: novel denoising methodology for brain MRI with adaptive masking and modified non-local block

Medical image denoising has been a subject of extensive research, with various techniques employed to enhance image quality and facilitate more accurate diagnostics. The evolution of denoising methods has highlighted impressive results but struggled to strike equilibrium between noise reduction and edge preservation which limits its applicability in various domains. This paper manifests the novel methodology that integrates an adaptive masking strategy, transformer-based U-Net Prior generator, edge enhancement module, and modified non-local block (MNLB) for denoising brain MRI clinical images. The adaptive masking strategy maintains the vital information through dynamic mask generation while the prior generator by capturing hierarchical features regenerates the high-quality prior MRI images. Finally, these images are fed to the edge enhancement module to boost structural information by maintaining crucial edge details, and the MNLB produces the denoised output by deriving non-local contextual information. The comprehensive experimental assessment is performed by employing two datasets namely the brain tumor MRI dataset and Alzheimer's dataset for diverse metrics and compared with conventional denoising approaches. The proposed denoising methodology achieves a PSNR of 40.965 and SSIM of 0.938 on the Alzheimer's dataset and also achieves a PSNR of 40.002 and SSIM of 0.926 on the brain tumor MRI dataset at a noise level of 50% revealing its supremacy in noise minimization. Furthermore, the impact of different masking ratios on denoising performance is analyzed which reveals that the proposed method showed PSNR of 40.965, SSIM of 0.938, MAE of 5.847, and MSE of 3.672 at the masking ratio of 60%. Moreover, the findings pave the way for the advancement of clinical image processing, facilitating precise detection of tumors in clinical MRI images.

Impact of Deep Learning Denoising Algorithm on Diffusion Tensor Imaging of the Growth Plate on Different Spatial Resolutions

To assess the impact of a deep learning (DL) denoising reconstruction algorithm applied to identical patient scans acquired with two different voxel dimensions, representing distinct spatial resolutions, this IRB-approved prospective study was conducted at a tertiary pediatric center in compliance with the Health Insurance Portability and Accountability Act. A General Electric Signa Premier unit (GE Medical Systems, Milwaukee, WI) was employed to acquire two DTI (diffusion tensor imaging) sequences of the left knee on each child at 3T: an in-plane 2.0 × 2.0 mm2 with section thickness of 3.0 mm and a 2 mm3 isovolumetric voxel; neither had an intersection gap. For image acquisition, a multi-band DTI with a fat-suppressed single-shot spin-echo echo-planar sequence (20 non-collinear directions; b-values of 0 and 600 s/mm2) was utilized. The MR vendor-provided a commercially available DL model which was applied with 75% noise reduction settings to the same subject DTI sequences at different spatial resolutions. We compared DTI tract metrics from both DL-reconstructed scans and non-denoised scans for the femur and tibia at each spatial resolution. Differences were evaluated using Wilcoxon-signed ranked test and Bland-Altman plots. When comparing DL versus non-denoised diffusion metrics in femur and tibia using the 2 mm × 2 mm × 3 mm voxel dimension, there were no significant differences between tract count (p = 0.1, p = 0.14) tract volume (p = 0.1, p = 0.29) or tibial tract length (p = 0.16); femur tract length exhibited a significant difference (p < 0.01). All diffusion metrics (tract count, volume, length, and fractional anisotropy (FA)) derived from the DL-reconstructed scans, were significantly different from the non-denoised scan DTI metrics in both the femur and tibial physes using the 2 mm3 voxel size (p < 0.001). DL reconstruction resulted in a significant decrease in femorotibial FA for both voxel dimensions (p < 0.01). Leveraging denoising algorithms could address the drawbacks of lower signal-to-noise ratios (SNRs) associated with smaller voxel volumes and capitalize on their better spatial resolutions, allowing for more accurate quantification of diffusion metrics.

Strategically Acquired Gradient Echo (STAGE) Imaging, part IV: Constrained Reconstruction of White Noise (CROWN) Processing as a Means to Improve Signal-to-Noise in STAGE Imaging at 3 Tesla

Increasing the signal-to-noise ratio (SNR) has always been of critical importance for magnetic resonance imaging. Although increasing field strength provides a linear increase in SNR, it is more and more costly as field strength increases. Therefore, there is a major effort today to use signal processing methods to improve SNR since it is more efficient and economical. There are a variety of methods to improve SNR such as averaging the data at the expense of imaging time, or collecting the data with a lower resolution, all of these methods, including imaging processing methods, usually come at the expense of loss of image detail or image blurring. Therefore, we developed a new mathematical approach called CROWN (Constrained Reconstruction of White Noise) to enhance SNR without loss of structural detail and without affecting scanning time. In this study, we introduced and tested the concept behind CROWN specifically for STAGE (strategically acquired gradient echo) imaging. The concept itself is presented first, followed by simulations to demonstrate its theoretical effectiveness. Then the SNR improvement on proton spin density (PSD) and R2⁎ maps was investigated using brain STAGE data acquired from 10 healthy controls (HCs) and 10 patients with Parkinson's disease (PD). For the PSD and R2* maps, the SNR and CNR between white matter and gray matter were improved by a factor of 1.87 ± 0.50 and 1.72 ± 0.88, respectively. The white matter hyperintensity lesions in PD patients were more clearly defined after CROWN processing. Using these improved maps, simulated images for any repeat time, echo time or flip angle can be created with improved SNR. The potential applications of this technology are to trade off the increased SNR for higher resolution images and/or faster imaging.

Romer-EPTI: rotating-view motion-robust super-resolution EPTI for SNR-efficient distortion-free in-vivo mesoscale dMRI and microstructure imaging

Purpose

To overcome the major challenges in dMRI acquisition, including low SNR, distortion/blurring, and motion vulnerability.

Methods

A novel Romer-EPTI technique is developed to provide distortion-free dMRI with significant SNR gain, high motion-robustness, sharp spatial resolution, and simultaneous multi-TE imaging. It introduces a ROtating-view Motion-robust supEr-Resolution technique (Romer) combined with a distortion/blurring-free EPTI encoding. Romer enhances SNR by a simultaneous multi-thick-slice acquisition with rotating-view encoding, while providing high motion-robustness through a motion-aware super-resolution reconstruction, which also incorporates slice-profile and real-value diffusion, to resolve high-isotropic-resolution volumes. The in-plane encoding is performed using distortion/blurring-free EPTI, which further improves effective spatial resolution and motion robustness by preventing not only T2/T2*-blurring but also additional blurring resulting from combining encoded volumes with inconsistent geometries caused by dynamic distortions. Self-navigation was incorporated to enable efficient phase correction. Additional developments include strategies to address slab-boundary artifacts, achieve minimal TE for SNR gain at 7T, and achieve high robustness to strong phase variations at high b-values.

Results

Using Romer-EPTI, we demonstrate distortion-free whole-brain mesoscale in-vivo dMRI at both 3T (500-μm-iso) and 7T (485-μm-iso) for the first time, with high SNR efficiency (e.g., 25 × ), and high image quality free from distortion and slab-boundary artifacts with minimal blurring. Motion experiments demonstrate Romer-EPTI's high motion-robustness and ability to recover sharp images in the presence of motion. Romer-EPTI also demonstrates significant SNR gain and robustness in high b-value (b=5000s/mm2) and time-dependent dMRI.

Conclusion

Romer-EPTI significantly improves SNR, motion-robustness, and image quality, providing a highly efficient acquisition for high-resolution dMRI and microstructure imaging.

High-angular resolution diffusion imaging generation using 3d u-net

PURPOSE

To investigate the effects on tractography of artificial intelligence-based prediction of motion-probing gradients (MPGs) in diffusion-weighted imaging (DWI).

METHODS

The 251 participants in this study were patients with brain tumors or epileptic seizures who underwent MRI to depict tractography. DWI was performed with 64 MPG directions and b = 0 s/mm2 images. The dataset was divided into a training set of 191 (mean age 45.7 [± 19.1] years), a validation set of 30 (mean age 41.6 [± 19.1] years), and a test set of 30 (mean age 49.6 [± 18.3] years) patients. Supervised training of a convolutional neural network was performed using b = 0 images and the first 32 axes of MPG images as the input data and the second 32 axes as the reference data. The trained model was applied to the test data, and tractography was performed using (a) input data only; (b) input plus prediction data; and (c) b = 0 images and the 64 MPG data (as a reference).

RESULTS

In Q-ball imaging tractography, the average dice similarity coefficient (DSC) of the input plus prediction data was 0.715 (± 0.064), which was significantly higher than that of the input data alone (0.697 [± 0.070]) (p < 0.05). In generalized q-sampling imaging tractography, the average DSC of the input plus prediction data was 0.769 (± 0.091), which was also significantly higher than that of the input data alone (0.738 [± 0.118]) (p < 0.01).

CONCLUSION

Diffusion tractography is improved by adding predicted MPG images generated by an artificial intelligence model.

MRI Assessment of Cerebral White Matter Microvascular Hemodynamics Across the Adult Lifespan

BACKGROUND

Changes in cerebral hemodynamics with aging are important for understanding age-related variation in neuronal health. While many prior studies have focused on gray matter, less is known regarding white matter due in part to measurement challenges related to the lower vascular density in white matter.

PURPOSE

To investigate the impact of age and sex on white matter hemodynamics in a Human Connectome Project in Aging (HCP-A) cohort using tract-based spatial statistics (TBSS).

STUDY TYPE

Retrospective cross-sectional.

POPULATION

Six hundred seventy-eight typically aging individuals (381 female), aged 36-100 years.

FIELD STRENGTH/SEQUENCE

Multi-delay pseudo-continuous arterial spin labeling (ASL) and diffusion-weighted pulsed-gradient spin-echo echo planar imaging sequences at 3.0 T.

ASSESSMENT

A skeleton of mean fractional anisotropy (FA) was produced using TBSS. This skeleton was used to project ASL-derived cerebral blood flow (CBF) and arterial transit time (ATT) measures onto white matter tracts.

STATISTICAL TESTS

General linear models were applied to white matter FA, CBF, and ATT maps, while covarying for age and sex. Threshold-free cluster enhancement multiple comparisons correction was performed for the effects of age and sex, thresholded at PFWE  < 0.05. CBF, ATT, and FA were compared between sex for each tract using analysis of covariance, with multiple comparisons correction for the number of tracts at PFDR  < 0.05.

RESULTS

Significantly lower white matter CBF and significantly prolonged white matter ATTs were associated with older age. These effects were widespread across tracts for ATT. Significant (PFDR  < 0.05) sex differences in ATT were observed across all tracts, and significant sex differences in CBF were observed in all tracts except the bilateral uncinate fasciculus. Females demonstrated significantly higher CBF compared to males across the lifespan. Few tracts demonstrated significant sex differences in FA.

DATA CONCLUSION

This study identified significant sex- and age-associated differences in white matter hemodynamics across tracts.

EVIDENCE LEVEL

3 TECHNICAL EFFICACY: Stage 3.

Structured low-rank reconstruction for navigator-free water/fat separated multi-shot diffusion-weighted EPI

PURPOSE

Multi-shot diffusion-weighted EPI allows an increase in image resolution and reduced geometric distortions and can be combined with chemical-shift encoding (Dixon) to separate water/fat signals. However, such approaches suffer from physiological motion-induced shot-to-shot phase variations. In this work, a structured low-rank-based navigator-free algorithm is proposed to address the challenge of simultaneously separating water/fat signals and correcting for physiological motion-induced shot-to-shot phase variations in multi-shot EPI-based diffusion-weighted MRI.

THEORY AND METHODS

We propose an iterative, model-based reconstruction pipeline that applies structured low-rank regularization to estimate and eliminate the shot-to-shot phase variations in a data-driven way, while separating water/fat images. The algorithm is tested in different anatomies, including head-neck, knee, brain, and prostate. The performance is validated in simulations and in-vivo experiments in comparison to existing approaches.

RESULTS

In-vivo experiments and simulations demonstrated the effectiveness of the proposed algorithm compared to extra-navigated and an alternative self-navigation approach. The proposed algorithm demonstrates the capability to reconstruct in the multi-shot/Dixon hybrid space domain under-sampled datasets, using the same number of acquired EPI shots compared to conventional fat-suppression techniques but eliminating fat signals through chemical-shift encoding. In addition, partial Fourier reconstruction can also be achieved by using the concept of virtual conjugate coils in conjunction with the proposed algorithm.

CONCLUSION

The proposed algorithm effectively eliminates the shot-to-shot phase variations and separates water/fat images, making it a promising solution for future DWI on different anatomies.

Multi-contrast high-field quality image synthesis for portable low-field MRI using generative adversarial networks and paired data

Introduction

Portable low-field strength (64mT) MRI scanners promise to increase access to neuroimaging for clinical and research purposes, however these devices produce lower quality images compared to high-field scanners. In this study, we developed and evaluated a deep learning architecture to generate high-field quality brain images from low-field inputs using a paired dataset of multiple sclerosis (MS) patients scanned at 64mT and 3T.

Methods

A total of 49 MS patients were scanned on portable 64mT and standard 3T scanners at Penn (n=25) or the National Institutes of Health (NIH, n=24) with T1-weighted, T2-weighted and FLAIR acquisitions. Using this paired data, we developed a generative adversarial network (GAN) architecture for low- to high-field image translation (LowGAN). We then evaluated synthesized images with respect to image quality, brain morphometry, and white matter lesions.

Results

Synthetic high-field images demonstrated visually superior quality compared to low-field inputs and significantly higher normalized cross-correlation (NCC) to actual high-field images for T1 (p=0.001) and FLAIR (p<0.001) contrasts. LowGAN generally outperformed the current state-of-the-art for low-field volumetrics. For example, thalamic, lateral ventricle, and total cortical volumes in LowGAN outputs did not differ significantly from 3T measurements. Synthetic outputs preserved MS lesions and captured a known inverse relationship between total lesion volume and thalamic volume.

Conclusions

LowGAN generates synthetic high-field images with comparable visual and quantitative quality to actual high-field scans. Enhancing portable MRI image quality could add value and boost clinician confidence, enabling wider adoption of this technology.

DENOISING DIFFUSION MRI: CONSIDERATIONS AND IMPLICATIONS FOR ANALYSIS

Development of diffusion MRI (dMRI) denoising approaches has experienced considerable growth over the last years. As noise can inherently reduce accuracy and precision in measurements, its effects have been well characterised both in terms of uncertainty increase in dMRI-derived features and in terms of biases caused by the noise floor, the smallest measurable signal given the noise level. However, gaps in our knowledge still exist in objectively characterising dMRI denoising approaches in terms of both of these effects and assessing their efficacy. In this work, we reconsider what a denoising method should and should not do and we accordingly define criteria to characterise the performance. We propose a comprehensive set of evaluations, including i) benefits in improving signal quality and reducing noise variance, ii) gains in reducing biases and the noise floor and improving, iii) preservation of spatial resolution, iv) agreement of denoised data against a gold standard, v) gains in downstream parameter estimation (precision and accuracy), vi) efficacy in enabling noise-prone applications, such as ultra-high-resolution imaging. We further provide newly acquired complex datasets (magnitude and phase) with multiple repeats that sample different SNR regimes to highlight performance differences under different scenarios. Without loss of generality, we subsequently apply a number of exemplar patch-based denoising algorithms to these datasets, including Non-Local Means, Marchenko-Pastur PCA (MPPCA) in the magnitude and complex domain and NORDIC, and compare them with respect to the above criteria and against a gold standard complex average of multiple repeats. We demonstrate that all tested denoising approaches reduce noise-related variance, but not always biases from the elevated noise floor. They all induce a spatial resolution penalty, but its extent can vary depending on the method and the implementation. Some denoising approaches agree with the gold standard more than others and we demonstrate challenges in even defining such a standard. Overall, we show that dMRI denoising performed in the complex domain is advantageous to magnitude domain denoising with respect to all the above criteria.

Low-rank tensor assisted K-space generative model for parallel imaging reconstruction

Although recent deep learning methods, especially generative models, have shown good performance in magnetic resonance imaging, there is still much room for improvement. Considering that the sample number and internal dimension in score-based generative model have key influence on estimating the gradients of data distribution, we present a new idea for parallel imaging reconstruction, named low-rank tensor assisted k-space generative model (LR-KGM). It means that we transform low-rank information into high-dimensional prior information for learning. More specifically, the multi-channel data is constructed into a large Hankel matrix to reduce the number of training samples, which is subsequently collapsed into a tensor for the stage of prior learning. In the testing phase, the low-rank rotation strategy is utilized to impose low-rank constraints on the output tensors of the generative network. Furthermore, we alternate the reconstruction between traditional generative iterations and low-rank high-dimensional tensor iterations. Experimental comparisons with the state-of-the-arts demonstrated that the proposed LR-KGM method achieved better performance.

SSL-QALAS: Self-Supervised Learning for rapid multiparameter estimation in quantitative MRI using 3D-QALAS

PURPOSE

To develop and evaluate a method for rapid estimation of multiparametric T1 , T2 , proton density, and inversion efficiency maps from 3D-quantification using an interleaved Look-Locker acquisition sequence with T2 preparation pulse (3D-QALAS) measurements using self-supervised learning (SSL) without the need for an external dictionary.

METHODS

An SSL-based QALAS mapping method (SSL-QALAS) was developed for rapid and dictionary-free estimation of multiparametric maps from 3D-QALAS measurements. The accuracy of the reconstructed quantitative maps using dictionary matching and SSL-QALAS was evaluated by comparing the estimated T1 and T2 values with those obtained from the reference methods on an International Society for Magnetic Resonance in Medicine/National Institute of Standards and Technology phantom. The SSL-QALAS and the dictionary-matching methods were also compared in vivo, and generalizability was evaluated by comparing the scan-specific, pre-trained, and transfer learning models.

RESULTS

Phantom experiments showed that both the dictionary-matching and SSL-QALAS methods produced T1 and T2 estimates that had a strong linear agreement with the reference values in the International Society for Magnetic Resonance in Medicine/National Institute of Standards and Technology phantom. Further, SSL-QALAS showed similar performance with dictionary matching in reconstructing the T1 , T2 , proton density, and inversion efficiency maps on in vivo data. Rapid reconstruction of multiparametric maps was enabled by inferring the data using a pre-trained SSL-QALAS model within 10 s. Fast scan-specific tuning was also demonstrated by fine-tuning the pre-trained model with the target subject's data within 15 min.

CONCLUSION

The proposed SSL-QALAS method enabled rapid reconstruction of multiparametric maps from 3D-QALAS measurements without an external dictionary or labeled ground-truth training data.

Learned spatiotemporal correlation priors for CEST image denoising using incorporated global-spectral convolution neural network

PURPOSE

To develop a deep learning-based method, dubbed Denoising CEST Network (DECENT), to fully exploit the spatiotemporal correlation prior to CEST image denoising.

METHODS

DECENT is composed of two parallel pathways with different convolution kernel sizes aiming to extract the global and spectral features embedded in CEST images. Each pathway consists of a modified U-Net with residual Encoder-Decoder network and 3D convolution. Fusion pathway with 1 × 1 × 1 convolution kernel is utilized to concatenate two parallel pathways, and the output of DECENT is noise-reduced CEST images. The performance of DECENT was validated in numerical simulations, egg white phantom experiments, and ischemic mouse brain and human skeletal muscle experiments in comparison with existing state-of-the-art denoising methods.

RESULTS

Rician noise was added to CEST images to mimic a low SNR situation for numerical simulation, egg white phantom experiment, and mouse brain experiments, while human skeletal muscle experiments were of inherently low SNR. From the denoising results evaluated by peak SNR (PSNR) and structural similarity index (SSIM), the proposed deep learning-based denoising method (DECENT) can achieve better performance compared to existing CEST denoising methods such as NLmCED, MLSVD, and BM4D, avoiding complicated parameter tuning or time-consuming iterative processes.

CONCLUSIONS

DECENT can well exploit the prior spatiotemporal correlation knowledge of CEST images and restore the noise-free images from their noisy observations, outperforming state-of-the-art denoising methods.

Deep Learning-Based Denoising of CEST MR Data: A Feasibility Study on Applying Synthetic Phantoms in Medical Imaging

Chemical Exchange Saturation Transfer (CEST) magnetic resonance imaging (MRI) provides a novel method for analyzing biomolecule concentrations in tissues without exogenous contrast agents. Despite its potential, achieving a high signal-to-noise ratio (SNR) is imperative for detecting small CEST effects. Traditional metrics such as Magnetization Transfer Ratio Asymmetry (MTRasym) and Lorentzian analyses are vulnerable to image noise, hampering their precision in quantitative concentration estimations. Recent noise-reduction algorithms like principal component analysis (PCA), nonlocal mean filtering (NLM), and block matching combined with 3D filtering (BM3D) have shown promise, as there is a burgeoning interest in the utilization of neural networks (NNs), particularly autoencoders, for imaging denoising. This study uses the Bloch-McConnell equations, which allow for the synthetic generation of CEST images and explores NNs efficacy in denoising these images. Using synthetically generated phantoms, autoencoders were created, and their performance was compared with traditional denoising methods using various datasets. The results underscored the superior performance of NNs, notably the ResUNet architectures, in noise identification and abatement compared to analytical approaches across a wide noise gamut. This superiority was particularly pronounced at elevated noise intensities in the in vitro data. Notably, the neural architectures significantly improved the PSNR values, achieving up to 35.0, while some traditional methods struggled, especially in low-noise reduction scenarios. However, the application to the in vivo data presented challenges due to varying noise profiles. This study accentuates the potential of NNs as robust denoising tools, but their translation to clinical settings warrants further investigation.

In vivo disentanglement of diffusion frequency-dependence, tensor shape, and relaxation using multidimensional MRI

Diffusion MRI with free gradient waveforms, combined with simultaneous relaxation encoding, referred to as multidimensional MRI (MD-MRI), offers microstructural specificity in complex biological tissue. This approach delivers intravoxel information about the microstructure, local chemical composition, and importantly, how these properties are coupled within heterogeneous tissue containing multiple microenvironments. Recent theoretical advances incorporated diffusion time dependency and integrated MD-MRI with concepts from oscillating gradients. This framework probes the diffusion frequency, ω , in addition to the diffusion tensor, D , and relaxation, R 1 , R 2 , correlations. A D ( ω ) - R 1 - R 2 clinical imaging protocol was then introduced, with limited brain coverage and 3 mm3 voxel size, which hinder brain segmentation and future cohort studies. In this study, we introduce an efficient, sparse in vivo MD-MRI acquisition protocol providing whole brain coverage at 2 mm3 voxel size. We demonstrate its feasibility and robustness using a well-defined phantom and repeated scans of five healthy individuals. Additionally, we test different denoising strategies to address the sparse nature of this protocol, and show that efficient MD-MRI encoding design demands a nuanced denoising approach. The MD-MRI framework provides rich information that allows resolving the diffusion frequency dependence into intravoxel components based on their D ( ω ) - R 1 - R 2 distribution, enabling the creation of microstructure-specific maps in the human brain. Our results encourage the broader adoption and use of this new imaging approach for characterizing healthy and pathological tissues.

Self-supervised structural similarity-based convolutional neural network for cardiac diffusion tensor image denoising

BACKGROUND

Diffusion tensor imaging (DTI) is a promising technique for non-invasively investigating the myocardial fiber structures of human heart. However, low signal-to-noise ratio (SNR) has been a major limit of cardiac DTI to prevent us from detecting myocardium structure accurately. Therefore, it is important to remove the effect of noise on diffusion weighted (DW) images.

PURPOSE

Although the conventional and deep learning-based denoising methods have shown the potential to deal with effectively the noise in DW images, most of them are redundant information dependent or require the noise-free images as golden standard. In addition, the existed DW image denoising methods often suffer from problems of over-smoothing. To address these issues, we propose a self-supervised learning model, structural similarity based convolutional neural network with edge-weighted loss (SSECNN), to remove the noise effectively in cardiac DTI.

METHODS

Considering that the DW images acquired along different diffusion directions have structural similarity, and the noise in these DW images is independent and identically distributed, the structural similarity-based matching algorithm is proposed to search for the most similar DW images. Such similar noisy DW image pairs are then used as the input and target of the denoising network SSECNN, which consists of several convolutional and residual blocks. Through the self-supervised training with these image pairs, the network can restore the clean DW images and retain the correlations between the denoised DW images along different directions. To avoid the over-smoothing problem, we design a novel edge-weighted loss which enables the network to adaptively adjust the loss weights with iterations and therefore to improve the detail preserve ability of the model. To verify the superiority of the proposed method, comparisons with state-of-the-art (SOTA) denoising methods are performed on both synthetic and real acquired DTI datasets.

RESULTS

Experimental results show that SSECNN can effectively reduce the noise in the DW images while preserving detailed texture and edge information and therefore achieve better performance in DTI reconstruction. For synthetic dataset, compared to the SOTA method, the root mean square error (RMSE), peak signal-to-noise ratio (PSNR), and structure similarity index measure (SSIM) between the denoised DW images obtained with SSECNN and noise-free DW images are improved by 6.94%, 1.98%, and 0.76% respectively when the noise level is 10%. As for the acquired cardiac DTI dataset, the SSECNN method could significantly improve SNR and contrast to noise ratio (CNR) of cardiac DW images and achieve more regular helix angle (HA) and transverse angle (TA) maps. The ablation experimental results validate that using the structure similarity-based method to search the similar DW image pairs yield the smallest loss, and with the help of the edge-weighted loss, the denoised DW images and diffusion metric maps can preserve more details.

CONCLUSIONS

The proposed SSECNN method can fully explore the similarity between the DW images along different diffusion directions. Using such similarity and an edge-weighted loss enable us to denoise cardiac DTI effectively in a self-supervised manner. Our method can overcome the redundancy information dependence and over-smoothing problem of the SOTA methods.

Diffusion MRI data analysis assisted by deep learning synthesized anatomical images (DeepAnat)

Diffusion MRI is a useful neuroimaging tool for non-invasive mapping of human brain microstructure and structural connections. The analysis of diffusion MRI data often requires brain segmentation, including volumetric segmentation and cerebral cortical surfaces, from additional high-resolution T1-weighted (T1w) anatomical MRI data, which may be unacquired, corrupted by subject motion or hardware failure, or cannot be accurately co-registered to the diffusion data that are not corrected for susceptibility-induced geometric distortion. To address these challenges, this study proposes to synthesize high-quality T1w anatomical images directly from diffusion data using convolutional neural networks (CNNs) (entitled "DeepAnat"), including a U-Net and a hybrid generative adversarial network (GAN), and perform brain segmentation on synthesized T1w images or assist the co-registration using synthesized T1w images. The quantitative and systematic evaluations using data of 60 young subjects provided by the Human Connectome Project (HCP) show that the synthesized T1w images and results for brain segmentation and comprehensive diffusion analysis tasks are highly similar to those from native T1w data. The brain segmentation accuracy is slightly higher for the U-Net than the GAN. The efficacy of DeepAnat is further validated on a larger dataset of 300 more elderly subjects provided by the UK Biobank. Moreover, the U-Nets trained and validated on the HCP and UK Biobank data are shown to be highly generalizable to the diffusion data from Massachusetts General Hospital Connectome Diffusion Microstructure Dataset (MGH CDMD) acquired with different hardware systems and imaging protocols and therefore can be used directly without retraining or with fine-tuning for further improved performance. Finally, it is quantitatively demonstrated that the alignment between native T1w images and diffusion images uncorrected for geometric distortion assisted by synthesized T1w images substantially improves upon that by directly co-registering the diffusion and T1w images using the data of 20 subjects from MGH CDMD. In summary, our study demonstrates the benefits and practical feasibility of DeepAnat for assisting various diffusion MRI data analyses and supports its use in neuroscientific applications.

Artificial intelligence for diffusion MRI-based tissue microstructure estimation in the human brain: an overview

Artificial intelligence (AI) has made significant advances in the field of diffusion magnetic resonance imaging (dMRI) and other neuroimaging modalities. These techniques have been applied to various areas such as image reconstruction, denoising, detecting and removing artifacts, segmentation, tissue microstructure modeling, brain connectivity analysis, and diagnosis support. State-of-the-art AI algorithms have the potential to leverage optimization techniques in dMRI to advance sensitivity and inference through biophysical models. While the use of AI in brain microstructures has the potential to revolutionize the way we study the brain and understand brain disorders, we need to be aware of the pitfalls and emerging best practices that can further advance this field. Additionally, since dMRI scans rely on sampling of the q-space geometry, it leaves room for creativity in data engineering in such a way that it maximizes the prior inference. Utilization of the inherent geometry has been shown to improve general inference quality and might be more reliable in identifying pathological differences. We acknowledge and classify AI-based approaches for dMRI using these unifying characteristics. This article also highlighted and reviewed general practices and pitfalls involving tissue microstructure estimation through data-driven techniques and provided directions for building on them.

Tensor denoising of multidimensional MRI data

PURPOSE

To develop a denoising strategy leveraging redundancy in high-dimensional data.

THEORY AND METHODS

The SNR fundamentally limits the information accessible by MRI. This limitation has been addressed by a host of denoising techniques, recently including the so-called MPPCA: principal component analysis of the signal followed by automated rank estimation, exploiting the Marchenko-Pastur distribution of noise singular values. Operating on matrices comprised of data patches, this popular approach objectively identifies noise components and, ideally, allows noise to be removed without introducing artifacts such as image blurring, or nonlocal averaging. The MPPCA rank estimation, however, relies on a large number of noise singular values relative to the number of signal components to avoid such ill effects. This condition is unlikely to be met when data patches and therefore matrices are small, for example due to spatially varying noise. Here, we introduce tensor MPPCA (tMPPCA) for the purpose of denoising multidimensional data, such as from multicontrast acquisitions. Rather than combining dimensions in matrices, tMPPCA uses each dimension of the multidimensional data's inherent tensor-structure to better characterize noise, and to recursively estimate signal components.

RESULTS

Relative to matrix-based MPPCA, tMPPCA requires no additional assumptions, and comparing the two in a numerical phantom and a multi-TE diffusion MRI data set, tMPPCA dramatically improves denoising performance. This is particularly true for small data patches, suggesting that tMPPCA can be especially beneficial in such cases.

CONCLUSIONS

The MPPCA denoising technique can be extended to high-dimensional data with improved performance for smaller patch sizes.

Deep Learning for Image Enhancement and Correction in Magnetic Resonance Imaging-State-of-the-Art and Challenges

Magnetic resonance imaging (MRI) provides excellent soft-tissue contrast for clinical diagnoses and research which underpin many recent breakthroughs in medicine and biology. The post-processing of reconstructed MR images is often automated for incorporation into MRI scanners by the manufacturers and increasingly plays a critical role in the final image quality for clinical reporting and interpretation. For image enhancement and correction, the post-processing steps include noise reduction, image artefact correction, and image resolution improvements. With the recent success of deep learning in many research fields, there is great potential to apply deep learning for MR image enhancement, and recent publications have demonstrated promising results. Motivated by the rapidly growing literature in this area, in this review paper, we provide a comprehensive overview of deep learning-based methods for post-processing MR images to enhance image quality and correct image artefacts. We aim to provide researchers in MRI or other research fields, including computer vision and image processing, a literature survey of deep learning approaches for MR image enhancement. We discuss the current limitations of the application of artificial intelligence in MRI and highlight possible directions for future developments. In the era of deep learning, we highlight the importance of a critical appraisal of the explanatory information provided and the generalizability of deep learning algorithms in medical imaging.

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

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

Multiple B-Value Model-Based Residual Network (MORN) for Accelerated High-Resolution Diffusion-Weighted Imaging

Single-Shot Echo Planar Imaging (SSEPI) based Diffusion Weighted Imaging (DWI) has shortcomings such as low resolution and severe distortions. In contrast, Multi-Shot EPI (MSEPI) provides optimal spatial resolution but increases scan time. This study proposed a Multiple b-value mOdel-based Residual Network (MORN) model to reconstruct multiple b-value high-resolution DWI from undersampled k-space data simultaneously. We incorporated Parallel Imaging (PI) into a residual U-net to reconstruct multiple b-value multi-coil data with the supervision of MUltiplexed Sensitivity-Encoding (MUSE) reconstructed Multi-Shot DWI (MSDWI). Moreover, asymmetric concatenations among different b-values and the combined loss to back propagate helped the feature transfer. After training and validation of the MORN in a dataset of 32 healthy cases, additional assessments were performed on 6 patients with different tumor types. The experimental results demonstrated that the MORN model outperformed conventional PI reconstruction (i.e. SENSE) and two state-of-the-art deep learning methods (SENSE-GAN and VSNet) in terms of PSNR (Peak Signal-to-Noise Ratio), SSIM (Structual SIMilarity) and apparent diffusion coefficient maps. In addition, using the pre-trained model under DWI, the MORN achieved consistent fractional anisotrophy and mean diffusivity reconstructed from multiple diffusion directions. Hence, the proposed method shows potential in clinical application according to the observations on tumor patients as well as images of multiple diffusion directions.

High B-value diffusion tensor imaging for early detection of hippocampal microstructural alteration in a mouse model of multiple sclerosis

Several studies have highlighted the value of diffusion tensor imaging (DTI) with strong diffusion weighting to reveal white matter microstructural lesions, but data in gray matter (GM) remains scarce. Herein, the effects of b-values combined with different numbers of diffusion-encoding directions (NDIRs) on DTI metrics to capture the normal hippocampal microstructure and its early alterations were investigated in a mouse model of multiple sclerosis (experimental autoimmune encephalomyelitis [EAE]). Two initial DTI datasets (B2700-43Dir acquired with b = 2700 s.mm⁻² and NDIR = 43; B1000-22Dir acquired with b = 1000 s.mm⁻² and NDIR = 22) were collected from 18 normal and 18 EAE mice at 4.7 T. Three additional datasets (B2700-22Dir, B2700-12Dir and B1000-12Dir) were extracted from the initial datasets. In healthy mice, we found a significant influence of b-values and NDIR on all DTI metrics. Confronting unsupervised hippocampal layers classification to the true anatomical classification highlighted the remarkable discrimination of the molecular layer with B2700-43Dir compared with the other datasets. Only DTI from the B2700 datasets captured the dendritic loss occurring in the molecular layer of EAE mice. Our findings stress the needs for both high b-values and sufficient NDIR to achieve a GM DTI with more biologically meaningful correlations, though DTI-metrics should be interpreted with caution in these settings.

A personalized deep learning denoising strategy for low-count PET images

Objective. Deep learning denoising networks are typically trained with images that are representative of the testing data. Due to the large variability of the noise levels in positron emission tomography (PET) images, it is challenging to develop a proper training set for general clinical use. Our work aims to develop a personalized denoising strategy for the low-count PET images at various noise levels.Approach.We first investigated the impact of the noise level in the training images on the model performance. Five 3D U-Net models were trained on five groups of images at different noise levels, and a one-size-fits-all model was trained on images covering a wider range of noise levels. We then developed a personalized weighting method by linearly blending the results from two models trained on 20%-count level images and 60%-count level images to balance the trade-off between noise reduction and spatial blurring. By adjusting the weighting factor, denoising can be conducted in a personalized and task-dependent way.Main results.The evaluation results of the six models showed that models trained on noisier images had better performance in denoising but introduced more spatial blurriness, and the one-size-fits-all model did not generalize well when deployed for testing images with a wide range of noise levels. The personalized denoising results showed that noisier images require higher weights on noise reduction to maximize the structural similarity and mean squared error. And model trained on 20%-count level images can produce the best liver lesion detectability.Significance.Our study demonstrated that in deep learning-based low dose PET denoising, noise levels in the training input images have a substantial impact on the model performance. The proposed personalized denoising strategy utilized two training sets to overcome the drawbacks introduced by each individual network and provided a series of denoised results for clinical reading.