q-Space Deep Learning: Twelve-Fold Shorter and Model-Free Diffusion MRI Scans

[Evaluation of Diffusional Kurtosis Inference Using Synthetic q-space Learning and Bias Correction]

PURPOSE

In synthetic q-space learning (synQSL), which uses deep learning to infer the diffusional kurtosis (K), a bias that depends on the noise level added to the synthetic training data occurs. The purpose of this study was to evaluate K inference using synQSL and bias correction.

METHODS

Using the synthetic test data and the real image data, K was inferred by synQSL, and bias correction was performed. Then, those results were compared with K inferred by fitting by the least-squares fitting (LSF) method. At this time, the noise level of the training data was set to 3 types, the noise level of the synthesis test data was set to 5 types, and the number of excitation (NEX) of the real image data was set to 4 types. Robustness of inference was evaluated by the outlier rate, which is the ratio of K outliers to the whole brain. We also evaluated the root mean square error (RMSE) of the inferred K.

RESULTS

The outlier rate inferred by synQSL without correction was significantly lower in the test data of each noise level than that by the LSF method and was further reduced by correction. In addition, the RMSE of NEX 1 with NEX 4 as the correct answer based on the real image data had the smallest correction result of K by synQSL.

CONCLUSION

Inferring K using synQSL and bias correction is a robust and small error method compared to that using the LSF method.

What's new and what's next in diffusion MRI preprocessing

Diffusion MRI (dMRI) provides invaluable information for the study of tissue microstructure and brain connectivity, but suffers from a range of imaging artifacts that greatly challenge the analysis of results and their interpretability if not appropriately accounted for. This review will cover dMRI artifacts and preprocessing steps, some of which have not typically been considered in existing pipelines or reviews, or have only gained attention in recent years: brain/skull extraction, B-matrix incompatibilities w.r.t the imaging data, signal drift, Gibbs ringing, noise distribution bias, denoising, between- and within-volumes motion, eddy currents, outliers, susceptibility distortions, EPI Nyquist ghosts, gradient deviations, B1 bias fields, and spatial normalization. The focus will be on "what's new" since the notable advances prior to and brought by the Human Connectome Project (HCP), as presented in the predecessing issue on "Mapping the Connectome" in 2013. In addition to the development of novel strategies for dMRI preprocessing, exciting progress has been made in the availability of open source tools and reproducible pipelines, databases and simulation tools for the evaluation of preprocessing steps, and automated quality control frameworks, amongst others. Finally, this review will consider practical considerations and our view on "what's next" in dMRI preprocessing.

Rapid MR relaxometry using deep learning: An overview of current techniques and emerging trends

Quantitative mapping of MR tissue parameters such as the spin-lattice relaxation time (T1 ), the spin-spin relaxation time (T2 ), and the spin-lattice relaxation in the rotating frame (T1ρ ), referred to as MR relaxometry in general, has demonstrated improved assessment in a wide range of clinical applications. Compared with conventional contrast-weighted (eg T1 -, T2 -, or T1ρ -weighted) MRI, MR relaxometry provides increased sensitivity to pathologies and delivers important information that can be more specific to tissue composition and microenvironment. The rise of deep learning in the past several years has been revolutionizing many aspects of MRI research, including image reconstruction, image analysis, and disease diagnosis and prognosis. Although deep learning has also shown great potential for MR relaxometry and quantitative MRI in general, this research direction has been much less explored to date. The goal of this paper is to discuss the applications of deep learning for rapid MR relaxometry and to review emerging deep-learning-based techniques that can be applied to improve MR relaxometry in terms of imaging speed, image quality, and quantification robustness. The paper is comprised of an introduction and four more sections. Section 2 describes a summary of the imaging models of quantitative MR relaxometry. In Section 3, we review existing "classical" methods for accelerating MR relaxometry, including state-of-the-art spatiotemporal acceleration techniques, model-based reconstruction methods, and efficient parameter generation approaches. Section 4 then presents how deep learning can be used to improve MR relaxometry and how it is linked to conventional techniques. The final section concludes the review by discussing the promise and existing challenges of deep learning for rapid MR relaxometry and potential solutions to address these challenges.

Multiparametric Functional MRI of the Kidney: Current State and Future Trends with Deep Learning Approaches

BACKGROUND

 Until today, assessment of renal function has remained a challenge for modern medicine. In many cases, kidney diseases accompanied by a decrease in renal function remain undetected and unsolved, since neither laboratory tests nor imaging diagnostics provide adequate information on kidney status. In recent years, developments in the field of functional magnetic resonance imaging with application to abdominal organs have opened new possibilities combining anatomic imaging with multiparametric functional information. The multiparametric approach enables the measurement of perfusion, diffusion, oxygenation, and tissue characterization in one examination, thus providing more comprehensive insight into pathophysiological processes of diseases as well as effects of therapeutic interventions. However, application of multiparametric fMRI in the kidneys is still restricted mainly to research areas and transfer to the clinical routine is still outstanding. One of the major challenges is the lack of a standardized protocol for acquisition and postprocessing including efficient strategies for data analysis. This article provides an overview of the most common fMRI techniques with application to the kidney together with new approaches regarding data analysis with deep learning.

METHODS

 This article implies a selective literature review using the literature database PubMed in May 2021 supplemented by our own experiences in this field.

RESULTS AND CONCLUSION

 Functional multiparametric MRI is a promising technique for assessing renal function in a more comprehensive approach by combining multiple parameters such as perfusion, diffusion, and BOLD imaging. New approaches with the application of deep learning techniques could substantially contribute to overcoming the challenge of handling the quantity of data and developing more efficient data postprocessing and analysis protocols. Thus, it can be hoped that multiparametric fMRI protocols can be sufficiently optimized to be used for routine renal examination and to assist clinicians in the diagnostics, monitoring, and treatment of kidney diseases in the future.

KEY POINTS

  · Multiparametric fMRI is a technique performed without the use of radiation, contrast media, and invasive methods.. · Multiparametric fMRI provides more comprehensive insight into pathophysiological processes of kidney diseases by combining functional and structural parameters.. · For broader acceptance of fMRI biomarkers, there is a need for standardization of acquisition, postprocessing, and analysis protocols as well as more prospective studies.. · Deep learning techniques could significantly contribute to an optimization of data acquisition and the postprocessing and interpretation of larger quantities of data..

CITATION FORMAT

· Zhang C, Schwartz M, Küstner T et al. Multiparametric Functional MRI of the Kidney: Current State and Future Trends with Deep Learning Approaches. Fortschr Röntgenstr 2022; DOI: 10.1055/a-1775-8633.

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

Diffusion tensor magnetic resonance imaging (DTI) is a widely adopted neuroimaging method for the in vivo mapping of brain tissue microstructure and white matter tracts. Nonetheless, the noise in the diffusion-weighted images (DWIs) decreases the accuracy and precision of DTI derived microstructural parameters and leads to prolonged acquisition time for achieving improved signal-to-noise ratio (SNR). Deep learning-based image denoising using convolutional neural networks (CNNs) has superior performance but often requires additional high-SNR data for supervising the training of CNNs, which reduces the feasibility of supervised learning-based denoising in practice. In this work, we develop a self-supervised deep learning-based method entitled "SDnDTI" for denoising DTI data, which does not require additional high-SNR data for training. Specifically, SDnDTI divides multi-directional DTI data into many subsets of six DWI volumes and transforms DWIs from each subset to along the same diffusion-encoding directions through the diffusion tensor model, generating multiple repetitions of DWIs with identical image contrasts but different noise observations. SDnDTI removes noise by first denoising each repetition of DWIs using a deep 3-dimensional CNN with the average of all repetitions with higher SNR as the training target, following the same approach as normal supervised learning based denoising methods, and then averaging CNN-denoised images for achieving higher SNR. The denoising efficacy of SDnDTI is demonstrated in terms of the similarity of output images and resultant DTI metrics compared to the ground truth generated using substantially more DWI volumes on two datasets with different spatial resolutions, b-values and numbers of input DWI volumes provided by the Human Connectome Project (HCP) and the Lifespan HCP in Aging. The SDnDTI results preserve image sharpness and textural details and substantially improve upon those from the raw data. The results of SDnDTI are comparable to those from supervised learning-based denoising and outperform those from state-of-the-art conventional denoising algorithms including BM4D, AONLM and MPPCA. By leveraging domain knowledge of diffusion MRI physics, SDnDTI makes it easier to use CNN-based denoising methods in practice and has the potential to benefit a wider range of research and clinical applications that require accelerated DTI acquisition and high-quality DTI data for mapping of tissue microstructure, fiber tracts and structural connectivity in the living human brain.

Contrast-Enhanced Fluid-Attenuated Inversion Recovery in Neuroimaging: A Narrative Review on Clinical Applications and Technical Advances

While contrast-enhanced fluid-attenuated inversion recovery (FLAIR) has long been regarded as an adjunct sequence to evaluate leptomeningeal disease in addition to contrast-enhanced T1-weighted imaging, it is gradually being used for more diverse pathologies beyond leptomeningeal disease. Contrast-enhanced FLAIR is known to be highly sensitive to low concentrations of gadolinium within the fluid. Accordingly, recent research has suggested the potential utility of contrast-enhanced FLAIR in various kinds of disease, such as Meniere's disease, seizure, stroke, traumatic brain injury, and brain metastasis, in addition to being used for visualizing glymphatic dysfunction. However, its potential applications have been reported sporadically in an unorganized manner. Furthermore, the exact mechanism for its superior sensitivity to low concentrations of gadolinium has not been fully understood. Rapidly developing magnetic resonance technology and unoptimized parameters for FLAIR may challenge its accurate application in clinical practice. This review provides the fundamental mechanism of contrast-enhanced FLAIR, systematically describes its current and potential clinical application, and elaborates on technical considerations for its optimization. LEVEL OF EVIDENCE: 3 TECHNICAL EFFICACY STAGE: 5.

Single-shell NODDI using dictionary-learner-estimated isotropic volume fraction

Neurite orientation dispersion and density imaging (NODDI) enables the assessment of intracellular, extracellular, and free water signals from multi-shell diffusion MRI data. It is an insightful approach to characterize brain tissue microstructure. Single-shell reconstruction for NODDI parameters has been discouraged in previous studies caused by failure when fitting, especially for the neurite density index (NDI). Here, we investigated the possibility of creating robust NODDI parameter maps with single-shell data, using the isotropic volume fraction (f_ISO ) as a prior. Prior estimation was made independent of the NODDI model constraint using a dictionary learning approach. First, we used a stochastic sparse dictionary-based network (DictNet), which is trained with data obtained from in vivo and simulated diffusion MRI data, to predict f_ISO . In single-shell cases, the mean diffusivity and raw T₂ signal with no diffusion weighting (S₀ ) was incorporated in the dictionary for the f_ISO estimation. Then, the NODDI framework was used with the known f_ISO to estimate the NDI and orientation dispersion index (ODI). The f_ISO estimated using our model was compared with other f_ISO estimators in the simulation. Further, using both synthetic data simulation and human data collected on a 3 T scanner (both high-quality HCP and clinical dataset), we compared the performance of our dictionary-based learning prior NODDI (DLpN) with the original NODDI for both single-shell and multi-shell data. Our results suggest that DLpN-derived NDI and ODI parameters for single-shell protocols are comparable with original multi-shell NODDI, and the protocol with b = 2000 s/mm² performs the best (error ~ 5% in white and gray matter). This may allow NODDI evaluation of studies on single-shell data by multi-shell scanning of two subjects for DictNet f_ISO training.

DIFFnet: Diffusion Parameter Mapping Network Generalized for Input Diffusion Gradient Schemes and b-Value

In MRI, deep neural networks have been proposed to reconstruct diffusion model parameters. However, the inputs of the networks were designed for a specific diffusion gradient scheme (i.e., diffusion gradient directions and numbers) and a specific b-value that are the same as the training data. In this study, a new deep neural network, referred to as DIFFnet, is developed to function as a generalized reconstruction tool of the diffusion-weighted signals for various gradient schemes and b-values. For generalization, diffusion signals are normalized in a q-space and then projected and quantized, producing a matrix (Qmatrix) as an input for the network. To demonstrate the validity of this approach, DIFFnet is evaluated for diffusion tensor imaging (DIFFnet_DTI) and for neurite orientation dispersion and density imaging (DIFFnet_NODDI). In each model, two datasets with different gradient schemes and b-values are tested. The results demonstrate accurate reconstruction of the diffusion parameters at substantially reduced processing time (approximately 8.7 times and 2240 times faster processing time than conventional methods in DTI and NODDI, respectively; less than 4% mean normalized root-mean-square errors (NRMSE) in DTI and less than 8% in NODDI). The generalization capability of the networks was further validated using reduced numbers of diffusion signals from the datasets and a public dataset from Human Connection Project. Different from previously proposed deep neural networks, DIFFnet does not require any specific gradient scheme and b-value for its input. As a result, it can be adopted as an online reconstruction tool for various complex diffusion imaging.

Jointly estimating parametric maps of multiple diffusion models from undersampled q-space data: A comparison of three deep learning approaches

PURPOSE

While advanced diffusion techniques have been found valuable in many studies, their clinical availability has been hampered partly due to their long scan times. Moreover, each diffusion technique can only extract a few relevant microstructural features. Using multiple diffusion methods may help to better understand the brain microstructure, which requires multiple expensive model fittings. In this work, we compare deep learning (DL) approaches to jointly estimate parametric maps of multiple diffusion representations/models from highly undersampled q-space data.

METHODS

We implement three DL approaches to jointly estimate parametric maps of diffusion tensor imaging (DTI), diffusion kurtosis imaging (DKI), neurite orientation dispersion and density imaging (NODDI), and multi-compartment spherical mean technique (SMT). A per-voxel q-space deep learning (1D-qDL), a per-slice convolutional neural network (2D-CNN), and a 3D-patch-based microstructure estimation with sparse coding using a separable dictionary (MESC-SD) network are considered.

RESULTS

The accuracy of estimated diffusion maps depends on the q-space undersampling, the selected network architecture, and the region and the parameter of interest. The smallest errors are observed for the MESC-SD network architecture (less than 10 % normalized RMSE in most brain regions).

CONCLUSION

Our experiments show that DL methods are very efficient tools to simultaneously estimate several diffusion maps from undersampled q-space data. These methods can significantly reduce both the scan ( ∼ 6-fold) and processing times ( ∼ 25-fold) for estimating advanced parametric diffusion maps while achieving a reasonable accuracy.

Single-shell to multi-shell dMRI transformation using spatial and volumetric multilevel hierarchical reconstruction framework

Single or Multi-shell high angular resolution diffusion imaging (HARDI) has become an important dMRI acquisition technique for studying brain white matter fibers. Existing single-shell HARDI makes it challenging to estimate the intravoxel structure up to the desired resolution. However, multi-shell acquisition (with multiple b-values) can provide higher resolution for the intravoxel structure, which further helps in getting accurate fiber tracts; But, this comes at the cost of larger acquisition time and larger setup. Hence, we propose a novel deep learning architecture for the reconstruction of diffusion MRI volumes for different b-values (degree of diffusion weighting) using acquisitions at a fixed b-value (termed as single-shell) acquisition. This reconstruction has been performed in the spherical harmonics space to better manage varying gradient directions. In this work, we have demonstrated such a reconstruction for b = 3000 s/mm² and b = 2000 s/mm² from b = 1000 s/mm². The proposed Multilevel Hierarchical Spherical Harmonics Coefficients Reconstruction (MHSH) framework takes advantage of contextual information within each slice as well as across the slices by involving Slice Level ReconNet (SLRNet) network and a Volumetric ROI Level ReconNet (VPLRNet) network, respectively. Three-loss functions have been used to optimize network learning, i.e., L₁, Adversarial, and Total Variation Loss. Finally, the network is trained and validated on the publicly available HCP data-set with standard qualitative and quantitative performance measures and achieves promising results.

Radiation Dose Reduction in Digital Mammography by Deep-Learning Algorithm Image Reconstruction: A Preliminary Study

Purpose

To develop a denoising convolutional neural network-based image processing technique and investigate its efficacy in diagnosing breast cancer using low-dose mammography imaging.

Materials and Methods

A total of 6 breast radiologists were included in this prospective study. All radiologists independently evaluated low-dose images for lesion detection and rated them for diagnostic quality using a qualitative scale. After application of the denoising network, the same radiologists evaluated lesion detectability and image quality. For clinical application, a consensus on lesion type and localization on preoperative mammographic examinations of breast cancer patients was reached after discussion. Thereafter, coded low-dose, reconstructed full-dose, and full-dose images were presented and assessed in a random order.

Results

Lesions on 40% reconstructed full-dose images were better perceived when compared with low-dose images of mastectomy specimens as a reference. In clinical application, as compared to 40% reconstructed images, higher values were given on full-dose images for resolution (p < 0.001); diagnostic quality for calcifications (p < 0.001); and for masses, asymmetry, or architectural distortion (p = 0.037). The 40% reconstructed images showed comparable values to 100% full-dose images for overall quality (p = 0.547), lesion visibility (p = 0.120), and contrast (p = 0.083), without significant differences.

Conclusion

Effective denoising and image reconstruction processing techniques can enable breast cancer diagnosis with substantial radiation dose reduction.

Prediction of dMRI signals with neural architecture search

BACKGROUND

There is growing interest in the neuroscience community in estimating and mapping microscopic properties of brain tissue non-invasively using magnetic resonance measurements. Machine learning methods are actively investigated to predict the signals measured in diffusion magnetic resonance imaging (dMRI).

NEW METHOD

We applied the neural architecture search (NAS) to train a recurrent neural network to generate a multilayer perceptron to predict the dMRI data of unknown signals based on the different acquisition parameters and training data. The search space of NAS is the number of neurons in each layer of the multilayer perceptron network. To our best knowledge, this is the first time to apply NAS to solve the dMRI signal prediction problem.

RESULTS

The experimental results demonstrate that the proposed NAS method can achieve fast training and predict dMRI signals accurately. For dMRI signals with four acquisition strategies of double diffusion encoding (DDE), double oscillating diffusion encoding (DODE), multi-shell and DSI-like pulsed gradient spin-echo (PGSE), the mean squared errors of the multilayer perceptron network designed by NAS are 0.0043, 0.0034, 0.0147 and 0.0199, respectively.

COMPARISON WITH EXISTING METHOD(S)

We also compared NAS with other machine learning prediction methods, such as support vector regression (SVR), decision tree (DT) and random forest (RF), k-nearest neighbors (KNN), adaboost regressor (AR), gradient boosting regressor (GBR) and extra-trees regressor (ET). NAS achieved the better prediction performance in most cases.

CONCLUSION

In this study, NAS was developed for the prediction of dMRI signals and could become an effective prediction tool.

High-fidelity approximation of grid- and shell-based sampling schemes from undersampled DSI using compressed sensing: Post mortem validation

While many useful microstructural indices, as well as orientation distribution functions, can be obtained from multi-shell dMRI data, there is growing interest in exploring the richer set of microstructural features that can be extracted from the full ensemble average propagator (EAP). The EAP can be readily computed from diffusion spectrum imaging (DSI) data, at the cost of a very lengthy acquisition. Compressed sensing (CS) has been used to make DSI more practical by reducing its acquisition time. CS applied to DSI (CS-DSI) attempts to reconstruct the EAP from significantly undersampled q-space data. We present a post mortem validation study where we evaluate the ability of CS-DSI to approximate not only fully sampled DSI but also multi-shell acquisitions with high fidelity. Human brain samples are imaged with high-resolution DSI at 9.4T and with polarization-sensitive optical coherence tomography (PSOCT). The latter provides direct measurements of axonal orientations at microscopic resolutions, allowing us to evaluate the mesoscopic orientation estimates obtained from diffusion MRI, in terms of their angular error and the presence of spurious peaks. We test two fast, dictionary-based, L2-regularized algorithms for CS-DSI reconstruction. We find that, for a CS acceleration factor of R=3, i.e., an acquisition with 171 gradient directions, one of these methods is able to achieve both low angular error and low number of spurious peaks. With a scan length similar to that of high angular resolution multi-shell acquisition schemes, this CS-DSI approach is able to approximate both fully sampled DSI and multi-shell data with high accuracy. Thus it is suitable for orientation reconstruction and microstructural modeling techniques that require either grid- or shell-based acquisitions. We find that the signal-to-noise ratio (SNR) of the training data used to construct the dictionary can have an impact on the accuracy of CS-DSI, but that there is substantial robustness to loss of SNR in the test data. Finally, we show that, as the CS acceleration factor increases beyond R=3, the accuracy of these reconstruction methods degrade, either in terms of the angular error, or in terms of the number of spurious peaks. Our results provide useful benchmarks for the future development of even more efficient q-space acceleration techniques.

Neural networks for parameter estimation in microstructural MRI: Application to a diffusion-relaxation model of white matter

Specific features of white matter microstructure can be investigated by using biophysical models to interpret relaxation-diffusion MRI brain data. Although more intricate models have the potential to reveal more details of the tissue, they also incur time-consuming parameter estimation that may converge to inaccurate solutions due to a prevalence of local minima in a degenerate fitting landscape. Machine-learning fitting algorithms have been proposed to accelerate the parameter estimation and increase the robustness of the attained estimates. So far, learning-based fitting approaches have been restricted to microstructural models with a reduced number of independent model parameters where dense sets of training data are easy to generate. Moreover, the degree to which machine learning can alleviate the degeneracy problem is poorly understood. For conventional least-squares solvers, it has been shown that degeneracy can be avoided by acquisition with optimized relaxation-diffusion-correlation protocols that include tensor-valued diffusion encoding. Whether machine-learning techniques can offset these acquisition requirements remains to be tested. In this work, we employ artificial neural networks to vastly accelerate the parameter estimation for a recently introduced relaxation-diffusion model of white matter microstructure. We also develop strategies for assessing the accuracy and sensitivity of function fitting networks and use those strategies to explore the impact of the acquisition protocol. The developed learning-based fitting pipelines were tested on relaxation-diffusion data acquired with optimal and sub-optimal acquisition protocols. Networks trained with an optimized protocol were observed to provide accurate parameter estimates within short computational times. Comparing neural networks and least-squares solvers, we found the performance of the former to be less affected by sub-optimal protocols; however, model fitting networks were still susceptible to degeneracy issues and their use could not fully replace a careful design of the acquisition protocol.

Deep-learning based super-resolution for 3D isotropic coronary MR angiography in less than a minute

PURPOSE

To develop and evaluate a novel and generalizable super-resolution (SR) deep-learning framework for motion-compensated isotropic 3D coronary MR angiography (CMRA), which allows free-breathing acquisitions in less than a minute.

METHODS

Undersampled motion-corrected reconstructions have enabled free-breathing isotropic 3D CMRA in ~5-10 min acquisition times. In this work, we propose a deep-learning-based SR framework, combined with non-rigid respiratory motion compensation, to shorten the acquisition time to less than 1 min. A generative adversarial network (GAN) is proposed consisting of two cascaded Enhanced Deep Residual Network generator, a trainable discriminator, and a perceptual loss network. A 16-fold increase in spatial resolution is achieved by reconstructing a high-resolution (HR) isotropic CMRA (0.9 mm3 or 1.2 mm3 ) from a low-resolution (LR) anisotropic CMRA (0.9 × 3.6 × 3.6 mm3 or 1.2 × 4.8 × 4.8 mm3 ). The impact and generalization of the proposed SRGAN approach to different input resolutions and operation on image and patch-level is investigated. SRGAN was evaluated on a retrospective downsampled cohort of 50 patients and on 16 prospective patients that were scanned with LR-CMRA in ~50 s under free-breathing. Vessel sharpness and length of the coronary arteries from the SR-CMRA is compared against the HR-CMRA.

RESULTS

SR-CMRA showed statistically significant (P < .001) improved vessel sharpness 34.1% ± 12.3% and length 41.5% ± 8.1% compared with LR-CMRA. Good generalization to input resolution and image/patch-level processing was found. SR-CMRA enabled recovery of coronary stenosis similar to HR-CMRA with comparable qualitative performance.

CONCLUSION

The proposed SR-CMRA provides a 16-fold increase in spatial resolution with comparable image quality to HR-CMRA while reducing the predictable scan time to <1 min.

Deep learning-based parameter estimation in fetal diffusion-weighted MRI

Diffusion-weighted magnetic resonance imaging (DW-MRI) of fetal brain is challenged by frequent fetal motion and signal to noise ratio that is much lower than non-fetal imaging. As a result, accurate and robust parameter estimation in fetal DW-MRI remains an open problem. Recently, deep learning techniques have been successfully used for DW-MRI parameter estimation in non-fetal subjects. However, none of those prior works has addressed the fetal brain because obtaining reliable fetal training data is challenging. To address this problem, in this work we propose a novel methodology that utilizes fetal scans as well as scans from prematurely-born infants. High-quality newborn scans are used to estimate accurate maps of the parameter of interest. These parameter maps are then used to generate DW-MRI data that match the measurement scheme and noise distribution that are characteristic of fetal data. In order to demonstrate the effectiveness and reliability of the proposed data generation pipeline, we used the generated data to train a convolutional neural network (CNN) to estimate color fractional anisotropy (CFA). We evaluated the trained CNN on independent sets of fetal data in terms of reconstruction accuracy, precision, and expert assessment of reconstruction quality. Results showed significantly lower reconstruction error (n=100,p<0.001) and higher reconstruction precision (n=20,p<0.001) for the proposed machine learning pipeline compared with standard estimation methods. Expert assessments on 20 fetal test scans showed significantly better overall reconstruction quality (p<0.001) and more accurate reconstruction of 11 regions of interest (p<0.001) with the proposed method.

On the generalizability of diffusion MRI signal representations across acquisition parameters, sequences and tissue types: Chronicles of the MEMENTO challenge

Diffusion MRI (dMRI) has become an invaluable tool to assess the microstructural organization of brain tissue. Depending on the specific acquisition settings, the dMRI signal encodes specific properties of the underlying diffusion process. In the last two decades, several signal representations have been proposed to fit the dMRI signal and decode such properties. Most methods, however, are tested and developed on a limited amount of data, and their applicability to other acquisition schemes remains unknown. With this work, we aimed to shed light on the generalizability of existing dMRI signal representations to different diffusion encoding parameters and brain tissue types. To this end, we organized a community challenge - named MEMENTO, making available the same datasets for fair comparisons across algorithms and techniques. We considered two state-of-the-art diffusion datasets, including single-diffusion-encoding (SDE) spin-echo data from a human brain with over 3820 unique diffusion weightings (the MASSIVE dataset), and double (oscillating) diffusion encoding data (DDE/DODE) of a mouse brain including over 2520 unique data points. A subset of the data sampled in 5 different voxels was openly distributed, and the challenge participants were asked to predict the remaining part of the data. After one year, eight participant teams submitted a total of 80 signal fits. For each submission, we evaluated the mean squared error, the variance of the prediction error and the Bayesian information criteria. The received submissions predicted either multi-shell SDE data (37%) or DODE data (22%), followed by cartesian SDE data (19%) and DDE (18%). Most submissions predicted the signals measured with SDE remarkably well, with the exception of low and very strong diffusion weightings. The prediction of DDE and DODE data seemed more challenging, likely because none of the submissions explicitly accounted for diffusion time and frequency. Next to the choice of the model, decisions on fit procedure and hyperparameters play a major role in the prediction performance, highlighting the importance of optimizing and reporting such choices. This work is a community effort to highlight strength and limitations of the field at representing dMRI acquired with trending encoding schemes, gaining insights into how different models generalize to different tissue types and fiber configurations over a large range of diffusion encodings.

Learning to estimate the fiber orientation distribution function from diffusion-weighted MRI

Estimation of white matter fiber orientation distribution function (fODF) is the essential first step for reliable brain tractography and connectivity analysis. Most of the existing fODF estimation methods rely on sub-optimal physical models of the diffusion signal or mathematical simplifications, which can impact the estimation accuracy. In this paper, we propose a data-driven method that avoids some of these pitfalls. Our proposed method is based on a multilayer perceptron that learns to map the diffusion-weighted measurements, interpolated onto a fixed spherical grid in the q space, to the target fODF. Importantly, we also propose methods for synthesizing reliable simulated training data. We show that the model can be effectively trained with simulated or real training data. Our phantom experiments show that the proposed method results in more accurate fODF estimation and tractography than several competing methods including the multi-tensor model, Bayesian estimation, spherical deconvolution, and two other machine learning techniques. On real data, we compare our method with other techniques in terms of accuracy of estimating the ground-truth fODF. The results show that our method is more accurate than other methods, and that it performs better than the competing methods when applied to under-sampled diffusion measurements. We also compare our method with the Sparse Fascicle Model in terms of expert ratings of the accuracy of reconstruction of several commissural, projection, association, and cerebellar tracts. The results show that the tracts reconstructed with the proposed method are rated significantly higher by three independent experts. Our study demonstrates the potential of data-driven methods for improving the accuracy and robustness of fODF estimation.

Training data distribution significantly impacts the estimation of tissue microstructure with machine learning

PURPOSE

Supervised machine learning (ML) provides a compelling alternative to traditional model fitting for parameter mapping in quantitative MRI. The aim of this work is to demonstrate and quantify the effect of different training data distributions on the accuracy and precision of parameter estimates when supervised ML is used for fitting.

METHODS

We fit a two- and three-compartment biophysical model to diffusion measurements from in-vivo human brain, as well as simulated diffusion data, using both traditional model fitting and supervised ML. For supervised ML, we train several artificial neural networks, as well as random forest regressors, on different distributions of ground truth parameters. We compare the accuracy and precision of parameter estimates obtained from the different estimation approaches using synthetic test data.

RESULTS

When the distribution of parameter combinations in the training set matches those observed in healthy human data sets, we observe high precision, but inaccurate estimates for atypical parameter combinations. In contrast, when training data is sampled uniformly from the entire plausible parameter space, estimates tend to be more accurate for atypical parameter combinations but may have lower precision for typical parameter combinations.

CONCLUSION

This work highlights that estimation of model parameters using supervised ML depends strongly on the training-set distribution. We show that high precision obtained using ML may mask strong bias, and visual assessment of the parameter maps is not sufficient for evaluating the quality of the estimates.

Noninterpretive Uses of Artificial Intelligence in Radiology

We deem a computer to exhibit artificial intelligence (AI) when it performs a task that would normally require intelligent action by a human. Much of the recent excitement about AI in the medical literature has revolved around the ability of AI models to recognize anatomy and detect pathology on medical images, sometimes at the level of expert physicians. However, AI can also be used to solve a wide range of noninterpretive problems that are relevant to radiologists and their patients. This review summarizes some of the newer noninterpretive uses of AI in radiology.

Q-space Conditioned Translation Networks for Directional Synthesis of Diffusion Weighted Images from Multi-modal Structural MRI

Current deep learning approaches for diffusion MRI modeling circumvent the need for densely-sampled diffusion-weighted images (DWIs) by directly predicting microstructural indices from sparsely-sampled DWIs. However, they implicitly make unrealistic assumptions of static q-space sampling during training and reconstruction. Further, such approaches can restrict downstream usage of variably sampled DWIs for usages including the estimation of microstructural indices or tractography. We propose a generative adversarial translation framework for high-quality DWI synthesis with arbitrary q-space sampling given commonly acquired structural images (e.g., B0, T1, T2). Our translation network linearly modulates its internal representations conditioned on continuous q-space information, thus removing the need for fixed sampling schemes. Moreover, this approach enables downstream estimation of high-quality microstructural maps from arbitrarily subsampled DWIs, which may be particularly important in cases with sparsely sampled DWIs. Across several recent methodologies, the proposed approach yields improved DWI synthesis accuracy and fidelity with enhanced downstream utility as quantified by the accuracy of scalar microstructure indices estimated from the synthesized images. Code is available at https://github.com/mengweiren/q-space-conditioned-dwi-synthesis.

A machine learning-based method for estimating the number and orientations of major fascicles in diffusion-weighted magnetic resonance imaging

Accurate modeling of diffusion-weighted magnetic resonance imaging measurements is necessary for accurate brain connectivity analysis. Existing methods for estimating the number and orientations of fascicles in an imaging voxel either depend on non-convex optimization techniques that are sensitive to initialization and measurement noise, or are prone to predicting spurious fascicles. In this paper, we propose a machine learning-based technique that can accurately estimate the number and orientations of fascicles in a voxel. Our method can be trained with either simulated or real diffusion-weighted imaging data. Our method estimates the angle to the closest fascicle for each direction in a set of discrete directions uniformly spread on the unit sphere. This information is then processed to extract the number and orientations of fascicles in a voxel. On realistic simulated phantom data with known ground truth, our method predicts the number and orientations of crossing fascicles more accurately than several classical and machine learning methods. It also leads to more accurate tractography. On real data, our method is better than or compares favorably with other methods in terms of robustness to measurement down-sampling and also in terms of expert quality assessment of tractography results.

Prospective Deployment of Deep Learning in MRI: A Framework for Important Considerations, Challenges, and Recommendations for Best Practices

Artificial intelligence algorithms based on principles of deep learning (DL) have made a large impact on the acquisition, reconstruction, and interpretation of MRI data. Despite the large number of retrospective studies using DL, there are fewer applications of DL in the clinic on a routine basis. To address this large translational gap, we review the recent publications to determine three major use cases that DL can have in MRI, namely, that of model-free image synthesis, model-based image reconstruction, and image or pixel-level classification. For each of these three areas, we provide a framework for important considerations that consist of appropriate model training paradigms, evaluation of model robustness, downstream clinical utility, opportunities for future advances, as well recommendations for best current practices. We draw inspiration for this framework from advances in computer vision in natural imaging as well as additional healthcare fields. We further emphasize the need for reproducibility of research studies through the sharing of datasets and software. LEVEL OF EVIDENCE: 5 TECHNICAL EFFICACY STAGE: 2.

qModeL: A plug-and-play model-based reconstruction for highly accelerated multi-shot diffusion MRI using learned priors

PURPOSE

To introduce a joint reconstruction method for highly undersampled multi-shot diffusion weighted (msDW) scans.

METHODS

Multi-shot EPI methods enable higher spatial resolution for diffusion MRI, but at the expense of long scan-time. Highly accelerated msDW scans are needed to enable their utilization in advanced microstructure studies, which require high q-space coverage. Previously, joint k-q undersampling methods coupled with compressed sensing were shown to enable very high acceleration factors. However, the reconstruction of this data using sparsity priors is challenging and is not suited for multi-shell data. We propose a new reconstruction that recovers images from the combined k-q data jointly. The proposed qModeL reconstruction brings together the advantages of model-based iterative reconstruction and machine learning, extending the idea of plug-and-play algorithms. Specifically, qModeL works by prelearning the signal manifold corresponding to the diffusion measurement space using deep learning. The prelearned manifold prior is incorporated into a model-based reconstruction to provide a voxel-wise regularization along the q-dimension during the joint recovery. Notably, the learning does not require in vivo training data and is derived exclusively from biophysical modeling. Additionally, a plug-and-play total variation denoising provides regularization along the spatial dimension. The proposed framework is tested on k-q undersampled single-shell and multi-shell msDW acquisition at various acceleration factors.

RESULTS

The qModeL joint reconstruction is shown to recover DWIs from 8-fold accelerated msDW acquisitions with error less than 5% for both single-shell and multi-shell data. Advanced microstructural analysis performed using the undersampled reconstruction also report reasonable accuracy.

CONCLUSION

qModeL enables the joint recovery of highly accelerated multi-shot dMRI utilizing learning-based priors. The bio-physically driven approach enables the use of accelerated multi-shot imaging for multi-shell sampling and advanced microstructure studies.

SuperDTI: Ultrafast DTI and fiber tractography with deep learning

PURPOSE

To develop a deep learning-based reconstruction framework for ultrafast and robust diffusion tensor imaging and fiber tractography.

METHODS

SuperDTI was developed to learn the nonlinear relationship between DWIs and the corresponding diffusion tensor parameter maps. It bypasses the tensor fitting procedure, which is highly susceptible to noises and motions in DWIs. The network was trained and tested using data sets from the Human Connectome Project and patients with ischemic stroke. Results from SuperDTI were compared against widely used methods for tensor parameter estimation and fiber tracking.

RESULTS

Using training and testing data acquired using the same protocol and scanner, SuperDTI was shown to generate fractional anisotropy and mean diffusivity maps, as well as fiber tractography, from as few as six raw DWIs, with a quantification error of less than 5% in all white-matter and gray-matter regions of interest. It was robust to noises and motions in the testing data. Furthermore, the network trained using healthy volunteer data showed no apparent reduction in lesion detectability when directly applied to stroke patient data.

CONCLUSIONS

Our results demonstrate the feasibility of superfast DTI and fiber tractography using deep learning with as few as six DWIs directly, bypassing tensor fitting. Such a significant reduction in scan time may allow the inclusion of DTI into the clinical routine for many potential applications.

Enhancing the estimation of fiber orientation distributions using convolutional neural networks

Local fiber orientation distributions (FODs) can be computed from diffusion magnetic resonance imaging (dMRI). The accuracy and ability of FODs to resolve complex fiber configurations benefits from acquisition protocols that sample a high number of gradient directions, a high maximum b-value, and multiple b-values. However, acquisition time and scanners that follow these standards are limited in clinical settings, often resulting in dMRI acquired at a single shell (single b-value). In this work, we learn improved FODs from clinically acquired dMRI. We evaluate patch-based 3D convolutional neural networks (CNNs) on their ability to regress multi-shell FODs from single-shell FODs, using constrained spherical deconvolution (CSD). We evaluate U-Net and High-Resolution Network (HighResNet) 3D CNN architectures on data from the Human Connectome Project and an in-house dataset. We evaluate how well each CNN can resolve FODs 1) when training and testing on datasets with the same dMRI acquisition protocol; 2) when testing on a dataset with a different dMRI acquisition protocol than used to train the CNN; and 3) when testing on a dataset with a fewer number of gradient directions than used to train the CNN. This work is a step towards more accurate FOD estimation in time- and resource-limited clinical environments.

Physics-based reconstruction methods for magnetic resonance imaging

Conventional magnetic resonance imaging (MRI) is hampered by long scan times and only qualitative image contrasts that prohibit a direct comparison between different systems. To address these limitations, model-based reconstructions explicitly model the physical laws that govern the MRI signal generation. By formulating image reconstruction as an inverse problem, quantitative maps of the underlying physical parameters can then be extracted directly from efficiently acquired k-space signals without intermediate image reconstruction-addressing both shortcomings of conventional MRI at the same time. This review will discuss basic concepts of model-based reconstructions and report on our experience in developing several model-based methods over the last decade using selected examples that are provided complete with data and code. This article is part of the theme issue 'Synergistic tomographic image reconstruction: part 1'.

On the potential for mapping apparent neural soma density via a clinically viable diffusion MRI protocol

•

B-tensor encoding enables estimation of spherical cellular structures in the brain.

•

Spherical compartments may provide markers for apparent neural soma density.

•

Model parameters can be estimated in a fast and robust way using deep learning.

•

Practical acquisition times are achievable on widely available clinical scanners.

Diffusion MRI is a valuable tool for probing tissue microstructure in the brain noninvasively. Today, model-based techniques are widely available and used for white matter characterisation where their development is relatively mature. Conversely, tissue modelling in grey matter is more challenging, and no generally accepted models exist. With advances in measurement technology and modelling efforts, a clinically viable technique that reveals salient features of grey matter microstructure, such as the density of quasi-spherical cell bodies and quasi-cylindrical cell projections, is an exciting prospect. As a step towards capturing the microscopic architecture of grey matter in clinically feasible settings, this work uses a biophysical model that is designed to disentangle the diffusion signatures of spherical and cylindrical structures in the presence of orientation heterogeneity, and takes advantage of B-tensor encoding measurements, which provide additional sensitivity compared to standard single diffusion encoding sequences. For the fast and robust estimation of microstructural parameters, we leverage recent advances in machine learning and replace conventional fitting techniques with an artificial neural network that fits complex biophysical models within seconds. Our results demonstrate apparent markers of spherical and cylindrical geometries in healthy human subjects, and in particular an increased volume fraction of spherical compartments in grey matter compared to white matter. We evaluate the extent to which spherical and cylindrical geometries may be interpreted as correlates of neural soma and neural projections, respectively, and quantify parameter estimation errors in the presence of various departures from the modelling assumptions. While further work is necessary to translate the ideas presented in this work to the clinic, we suggest that biomarkers focussing on quasi-spherical cellular geometries may be valuable for the enhanced assessment of neurodevelopmental disorders and neurodegenerative diseases.

Deep learning based segmentation of brain tissue from diffusion MRI

Segmentation of brain tissue types from diffusion MRI (dMRI) is an important task, required for quantification of brain microstructure and for improving tractography. Current dMRI segmentation is mostly based on anatomical MRI (e.g., T1- and T2-weighted) segmentation that is registered to the dMRI space. However, such inter-modality registration is challenging due to more image distortions and lower image resolution in dMRI as compared with anatomical MRI. In this study, we present a deep learning method for diffusion MRI segmentation, which we refer to as DDSeg. Our proposed method learns tissue segmentation from high-quality imaging data from the Human Connectome Project (HCP), where registration of anatomical MRI to dMRI is more precise. The method is then able to predict a tissue segmentation directly from new dMRI data, including data collected with different acquisition protocols, without requiring anatomical data and inter-modality registration. We train a convolutional neural network (CNN) to learn a tissue segmentation model using a novel augmented target loss function designed to improve accuracy in regions of tissue boundary. To further improve accuracy, our method adds diffusion kurtosis imaging (DKI) parameters that characterize non-Gaussian water molecule diffusion to the conventional diffusion tensor imaging parameters. The DKI parameters are calculated from the recently proposed mean-kurtosis-curve method that corrects implausible DKI parameter values and provides additional features that discriminate between tissue types. We demonstrate high tissue segmentation accuracy on HCP data, and also when applying the HCP-trained model on dMRI data from other acquisitions with lower resolution and fewer gradient directions.

Recent Advances in Parameter Inference for Diffusion MRI Signal Models

In this paper, fundamentals and recent progress for obtaining biological features quantitatively by using diffusion MRI are reviewed. First, a brief description of diffusion MRI history, application, and development was presented. Then, well-known parametric models including diffusion tensor imaging (DTI), diffusional kurtosis imaging (DKI), and neurite orientation dispersion diffusion imaging (NODDI) are introduced with several classifications in various viewpoints with other modeling schemes. In addition, this review covers mathematical generalization and examples of methodologies for the model parameter inference from conventional fitting to recent machine learning approaches, which is called Q-space learning (QSL). Finally, future perspectives on diffusion MRI parameter inference are discussed with the aspects of imaging modeling and simulation.

Improving phase-based conductivity reconstruction by means of deep learning-based denoising of B 1 + phase data for 3T MRI

PURPOSE

To denoise B 1 + phase using a deep learning method for phase-based in vivo electrical conductivity reconstruction in a 3T MR system.

METHODS

For B 1 + phase deep-learning denoising, a convolutional neural network (U-net) was chosen. Training was performed on data sets from 10 healthy volunteers. Input data were the real and imaginary components of single averaged spin-echo data (SNR = 45), which was used to approximate the B 1 + phase. For label data, multiple signal-averaged spin-echo data (SNR = 128) were used. Testing was performed on in silico and in vivo data. Reconstructed conductivity maps were derived using phase-based conductivity reconstructions. Additionally, we investigated the usability of the network to various SNR levels, imaging contrasts, and anatomical sites (ie, T₁ , T₂ , and proton density-weighted brain images and proton density-weighted breast images. In addition, conductivity reconstructions from deep learning-based denoised data were compared with conventional image filters, which were used for data denoising in electrical properties tomography (ie, the Gaussian filtering and the Savitzky-Golay filtering).

RESULTS

The proposed deep learning-based denoising approach showed improvement for B 1 + phase for both in silico and in vivo experiments with reduced quantitative error measures compared with other methods. Subsequently, this resulted in an improvement of reconstructed conductivity maps from the denoised B 1 + phase with deep learning.

CONCLUSION

The results suggest that the proposed approach can be used as an alternative preprocessing method to denoise B 1 + maps for phase-based conductivity reconstruction without relying on image filters or signal averaging.

Multimodal super-resolved q-space deep learning

Super-resolvedq-space deep learning (SR-q-DL) has been developed to estimate high-resolution (HR) tissue microstructure maps from low-quality diffusion magnetic resonance imaging (dMRI) scans acquired with a reduced number of diffusion gradients and low spatial resolution, where deep networks are designed for the estimation. However, existing methods do not exploit HR information from other modalities, which are generally acquired together with dMRI and could provide additional useful information for HR tissue microstructure estimation. In this work, we extend SR-q-DL and propose multimodal SR-q-DL, where information in low-resolution (LR) dMRI is combined with HR information from another modality for HR tissue microstructure estimation. Because the HR modality may not be as sensitive to tissue microstructure as dMRI, direct concatenation of multimodal information does not necessarily lead to improved estimation performance. Since existing deep networks for HR tissue microstructure estimation are patch-based and use redundant information in the spatial domain to enhance the spatial resolution, the HR information in the other modality could inform the deep networks about what input voxels are relevant for the computation of tissue microstructure. Thus, we propose to incorporate the HR information from the HR modality by designing an attention module that guides the computation of HR tissue microstructure from LR dMRI. Specifically, the attention module is integrated with the patch-based SR-q-DL framework that exploits the sparsity of diffusion signals. The sparse representation of the LR diffusion signals in the input patch is first computed with a network component that unrolls an iterative process for sparse reconstruction. Then, the proposed attention module computes a relevance map from the HR modality with sequential convolutional layers. The relevance map indicates the relevance of the LR sparse representation at each voxel for computing the patch of HR tissue microstructure. The relevance is applied to the LR sparse representation with voxelwise multiplication, and the weighted LR sparse representation is used to compute HR tissue microstructure with another network component that allows resolution enhancement. All weights in the proposed network for multimodal SR-q-DL are jointly learned and the estimation is end-to-end. To evaluate the proposed method, we performed experiments on brain dMRI scans together with images of additional HR modalities. In the experiments, the proposed method was applied to the estimation of tissue microstructure measures for different datasets and advanced biophysical models, where the benefit of incorporating multimodal information using the proposed method is shown.

Artificial Intelligence for MR Image Reconstruction: An Overview for Clinicians

Artificial intelligence (AI) shows tremendous promise in the field of medical imaging, with recent breakthroughs applying deep-learning models for data acquisition, classification problems, segmentation, image synthesis, and image reconstruction. With an eye towards clinical applications, we summarize the active field of deep-learning-based MR image reconstruction. We review the basic concepts of how deep-learning algorithms aid in the transformation of raw k-space data to image data, and specifically examine accelerated imaging and artifact suppression. Recent efforts in these areas show that deep-learning-based algorithms can match and, in some cases, eclipse conventional reconstruction methods in terms of image quality and computational efficiency across a host of clinical imaging applications, including musculoskeletal, abdominal, cardiac, and brain imaging. This article is an introductory overview aimed at clinical radiologists with no experience in deep-learning-based MR image reconstruction and should enable them to understand the basic concepts and current clinical applications of this rapidly growing area of research across multiple organ systems.

Use of deep learning for detection, characterisation and prediction of metastatic disease from computerised tomography: a systematic review

CT is widely used for diagnosis, staging and management of cancer. The presence of metastasis has significant implications on treatment and prognosis. Deep learning (DL), a form of machine learning, where layers of programmed algorithms interpret and recognise patterns, may have a potential role in CT image analysis. This review aims to provide an overview on the use of DL in CT image analysis in the diagnostic evaluation of metastatic disease. A total of 29 studies were included which could be grouped together into three areas of research: the use of deep learning on the detection of metastatic disease from CT imaging, characterisation of lesions on CT into metastasis and prediction of the presence or development of metastasis based on the primary tumour. In conclusion, DL in CT image analysis could have a potential role in evaluating metastatic disease; however, prospective clinical trials investigating its clinical value are required.

Quantification of intravoxel incoherent motion with optimized b-values using deep neural network

PURPOSE

To develop a framework for quantifying intravoxel incoherent motion (IVIM) parameters, where a neural network for quantification and b-values for diffusion-weighted imaging are simultaneously optimized.

METHOD

A deep neural network (DNN) method is proposed for accurate quantification of IVIM parameters from multiple diffusion-weighted images. In addition, optimal b-values are selected to acquire the multiple diffusion-weighted images. The proposed framework consists of an MRI signal generation part and an IVIM parameter quantification part. Monte-Carlo (MC) simulations were performed to evaluate the accuracy of the IVIM parameter quantification and the efficacy of b-value optimization. In order to analyze the effect of noise on the optimized b-values, simulations were performed with five different noise levels. For in vivo data, diffusion images were acquired with the b-values from four b-values selection methods for five healthy volunteers at 3T MRI system.

RESULTS

Experiment results showed that both the optimization of b-values and the training of DNN were simultaneously performed to quantify IVIM parameters. We found that the accuracies of the perfusion coefficient (D_p ) and perfusion fraction (f) were more sensitive to b-values than the diffusion coefficient (D) was. Furthermore, when the noise level changed, the optimized b-values also changed. Therefore, noise level has to be considered when optimizing b-values for IVIM quantification.

CONCLUSION

The proposed scheme can simultaneously optimize b-values and train DNN to minimize quantification errors of IVIM parameters. The trained DNN can quantify IVIM parameters from the diffusion-weighted images obtained with the optimized b-values.

Simultaneous multi-slice image reconstruction using regularized image domain split slice-GRAPPA for diffusion MRI

The main goal of this work is to improve the quality of simultaneous multi-slice (SMS) reconstruction for diffusion MRI. We accomplish this by developing an image domain method that reaps the benefits of both SENSE and GRAPPA-type approaches and enables image regularization in an optimization framework. We propose a new approach termed regularized image domain split slice-GRAPPA (RI-SSG), which establishes an optimization framework for SMS reconstruction. Within this framework, we use a robust forward model to take advantage of both the SENSE model with explicit sensitivity estimations and the SSG model with implicit kernel relationship among coil images. The proposed approach also allows combining of coil images to increase the SNR and enables image domain regularization on estimated coil-combined single slices. We compare the performance of RI-SSG with that of SENSE and SSG using in-vivo diffusion EPI datasets with simulated and actual SMS acquisitions collected on a 3T MR scanner. Reconstructed diffusion-weighted images (DWIs) and the resulting diffusion tensor imaging (DTI) and neurite orientation dispersion and density imaging (NODDI) maps are analyzed to evaluate the quantitative and qualitative performance of the three methods. The DWIs reconstructed by RI-SSG are closer to the single-band ground truth images than SENSE and SSG. Specifically, the proposed RI-SSG reduces the normalized root-mean-square-error (nRMSE) against ground truth images by ∼5% and increases the structural similarity index (SSIM) by ∼4% compared to SSG. All three methods produce similar fractional anisotropy (FA) maps using DTI representation, but mean diffusivity (MD) and fiber orientation estimates using RI-SSG are closer to the reference than SENSE and SSG. RI-SSG results in NODDI maps with noticeably smaller errors than those of SENSE and SSG and improves the accuracy of the mean value of orientation dispersion index (ODI) by ∼5% and the mean value of intracellular volume fraction by ∼7% in regions of interest in brain white matter compared to SSG.

Reducing scan time of paediatric 99mTc-DMSA SPECT via deep learning

AIM

To investigate the feasibility of reducing the scan time of paediatric technetium 99m (^99mTc) dimercaptosuccinic acid (DMSA) single-photon-emission computed tomographic (SPECT) using a deep learning (DL) method.

MATERIAL AND METHODS

A total of 112 paediatric ^99mTc-DMSA renal SPECT scans were analysed retrospectively. Of the 112 examinations, 88 (84 for training and four for validation) were used to train a DL-based model that could generate full-acquisition-time reconstructed SPECT images from half-time acquisition. The remaining 24 examinations were used to evaluate the performance of the trained model.

RESULTS

DL-based SPECT images obtained from half-time acquisition have image quality similar to the standard clinical SPECT images obtained from full-acquisition-time acquisition. Moreover, the accuracy, sensitivity and specificity of the DL-based SPECT images for detection of affected kidneys were 91.7%, 83.3%, and 100%, respectively.

CONCLUSION

These preliminary results suggest that DL has the potential to reduce the scan time of paediatric ^99mTc-DMSA SPECT imaging while maintaining diagnostic accuracy.

Denouements of machine learning and multimodal diagnostic classification of Alzheimer's disease

Alzheimer's disease (AD) is the most common type of dementia. The exact cause and treatment of the disease are still unknown. Different neuroimaging modalities, such as magnetic resonance imaging (MRI), positron emission tomography, and single-photon emission computed tomography, have played a significant role in the study of AD. However, the effective diagnosis of AD, as well as mild cognitive impairment (MCI), has recently drawn large attention. Various technological advancements, such as robots, global positioning system technology, sensors, and machine learning (ML) algorithms, have helped improve the diagnostic process of AD. This study aimed to determine the influence of implementing different ML classifiers in MRI and analyze the use of support vector machines with various multimodal scans for classifying patients with AD/MCI and healthy controls. Conclusions have been drawn in terms of employing different classifier techniques and presenting the optimal multimodal paradigm for the classification of AD.

Super-Resolved q-Space deep learning with uncertainty quantification

Diffusion magnetic resonance imaging (dMRI) provides a noninvasive method for measuring brain tissue microstructure. q-Space deep learning(q-DL) methods have been developed to accurately estimate tissue microstructure from dMRI scans acquired with a reduced number of diffusion gradients. In these methods, deep networks are trained to learn the mapping directly from diffusion signals to tissue microstructure. However, the quality of tissue microstructure estimation can be limited not only by the reduced number of diffusion gradients but also by the low spatial resolution of typical dMRI acquisitions. Therefore, in this work we extend q-DL to super-resolved tissue microstructure estimation and propose super-resolvedq-DL (SR-q-DL), where deep networks are designed to map low-resolution diffusion signals undersampled in the q-space to high-resolution tissue microstructure. Specifically, we use a patch-based strategy, where a deep network takes low-resolution patches of diffusion signals as input and outputs high-resolution tissue microstructure patches. The high-resolution patches are then combined to obtain the final high-resolution tissue microstructure map. Motivated by existing q-DL methods, we integrate the sparsity of diffusion signals in the network design, which comprises two functional components. The first component computes sparse representation of diffusion signals for the low-resolution input patch, and the second component maps the low-resolution sparse representation to high-resolution tissue microstructure. The weights in the two components are learned jointly and the trained network performs end-to-end tissue microstructure estimation. In addition to SR-q-DL, we further propose probabilistic SR-q-DL, which can quantify the uncertainty of the network output as well as achieve improved estimation accuracy. In probabilistic SR-q-DL, a deep ensemble strategy is used. Specifically, the deep network for SR-q-DL is revised to produce not only tissue microstructure estimates but also the uncertainty of the estimates. Then, multiple deep networks are trained and their results are fused for the final prediction of high-resolution tissue microstructure and uncertainty quantification. The proposed method was evaluated on two independent datasets of brain dMRI scans. Results indicate that our approach outperforms competing methods in terms of estimation accuracy. In addition, uncertainty measures provided by our method correlate with estimation errors, which indicates potential application of the proposed uncertainty quantification method in brain studies.

Multi-Shell D-MRI Reconstruction via Residual Learning utilizing Encoder-Decoder Network with Attention (MSR-Net)

Contemporary diffusion MRI based analysis with HARDI, which provides more accurate fiber orientation, can be performed using single or multiple b-values (single or multi-shell). Single shell HARDI cannot provide volume fraction for different tissue types, which can produce bias and noisier results in estimation of fiber ODF. Multi-shell acquisition can resolve this issue. However, it requires more scanning time and is therefore not very well suited in clinical setting. Considering this, we propose a novel deep learning architecture, MSR-Net, for reconstruction of diffusion MRI volumes for some b-value using acquisitions at another b-value. In this work, we demonstrate this for b = 2000 s/mm² and b = 1000 s/mm². We learn such a transformation in the space of spherical harmonic coefficients. The proposed network consists of encoder-decoder along-with an attention module and a feature module. We have considered L2 and Content loss for optimizing and improving the performance. We have trained and validated the network using the HCP data-set with standard qualitative and quantitative performance measures.

Deep learning-based method for reducing residual motion effects in diffusion parameter estimation

PURPOSE

Conventional motion-correction techniques for diffusion MRI can introduce motion-level-dependent bias in derived metrics. To address this challenge, a deep learning-based technique was developed to minimize such residual motion effects.

METHODS

The data-rejection approach was adopted in which motion-corrupted data are discarded before model-fitting. A deep learning-based parameter estimation algorithm, using a hierarchical convolutional neural network (H-CNN), was combined with motion assessment and corrupted volume rejection. The method was designed to overcome the limitations of existing methods of this kind that produce parameter estimations whose quality depends strongly on a proportion of the data discarded. Evaluation experiments were conducted for the estimation of diffusion kurtosis and diffusion-tensor-derived measures at both the individual and group levels. The performance was compared with the robust approach of iteratively reweighted linear least squares (IRLLS) after motion correction with and without outlier replacement.

RESULTS

Compared with IRLLS, the H-CNN-based technique is minimally sensitive to motion effects. It was tested at severe motion levels when 70% to 90% of the data are rejected and when random motion is present. The technique had a stable performance independent of the numbers and schemes of data rejection. A further test on a data set from children with attention-deficit hyperactivity disorder shows the technique can potentially ameliorate spurious group-level difference caused by head motion.

CONCLUSION

This method shows great potential for reducing residual motion effects in motion-corrupted diffusion-weighted-imaging data, bringing benefits that include reduced bias in derived metrics in individual scans and reduced motion-level-dependent bias in population studies employing diffusion MRI.

Accelerating in vivo fast spin echo high angular resolution diffusion imaging with an isotropic resolution in mice through compressed sensing

PURPOSE

Echo planar imaging (EPI) is commonly used to acquire the many volumes needed for high angular resolution diffusion Imaging (HARDI), posing a higher risk for artifacts, such as distortion and deformation. An alternative to EPI is fast spin echo (FSE) imaging, which has fewer artifacts but is inherently slower. The aim is to accelerate FSE such that a HARDI data set can be acquired in a time comparable to EPI using compressed sensing.

METHODS

Compressed sensing was applied in either q-space or simultaneously in k-space and q-space, by undersampling the k-space in the phase-encoding direction or retrospectively eliminating diffusion directions for different degrees of undersampling. To test the replicability of the acquisition and reconstruction, brain data were acquired from six mice, and a numerical phantom experiment was performed. All HARDI data were analyzed individually using constrained spherical deconvolution, and the apparent fiber density and complexity metric were evaluated, together with whole-brain tractography.

RESULTS

The apparent fiber density and complexity metric showed relatively minor differences when only q-space undersampling was used, but deteriorate when k-space undersampling was applied. Likewise, the tract density weighted image showed good results when only q-space undersampling was applied using 15 directions or more, but information was lost when fewer volumes or k-space undersampling were used.

CONCLUSION

It was found that acquiring 15 to 20 diffusion directions with a full k-space and reconstructed using compressed sensing could suffice for a replicable measurement of quantitative measures in mice, where areas near the sinuses and ear cavities are untainted by signal loss.

Challenges for biophysical modeling of microstructure

The biophysical modeling efforts in diffusion MRI have grown considerably over the past 25 years. In this review, we dwell on the various challenges along the journey of bringing a biophysical model from initial design to clinical implementation, identifying both hurdles that have been already overcome and outstanding issues. First, we describe the critical initial task of selecting which features of tissue microstructure can be estimated using a model and which acquisition protocol needs to be implemented to make the estimation possible. The model performance should necessarily be tested in realistic numerical simulations and in experimental data - adapting the fitting strategy accordingly, and parameter estimates should be validated against complementary techniques, when/if available. Secondly, the model performance and validity should be explored in pathological conditions, and, if appropriate, dedicated models for pathology should be developed. We build on examples from tumors, ischemia and demyelinating diseases. We then discuss the challenges associated with clinical translation and added value. Finally, we single out four major unresolved challenges that are related to: the availability of a microstructural ground truth, the validation of model parameters which cannot be accessed with complementary techniques, the development of a generalized standard model for any brain region and pathology, and the seamless communication between different parties involved in the development and application of biophysical models of diffusion.

Estimating Tissue Microstructure with Undersampled Diffusion Data via Graph Convolutional Neural Networks

Advanced diffusion models for tissue microstructure are widely employed to study brain disorders. However, these models usually require diffusion MRI (DMRI) data with densely sampled q-space, which is prohibitive in clinical settings. This problem can be resolved by using deep learning techniques, which learn the mapping between sparsely sampled q-space data and the high-quality diffusion microstructural indices estimated from densely sampled data. However, most existing methods simply view the input DMRI data as a vector without considering data structure in the q-space. In this paper, we propose to overcome this limitation by representing DMRI data using graphs and utilizing graph convolutional neural networks to estimate tissue microstructure. Our method makes full use of the q-space angular neighboring information to improve estimation accuracy. Experimental results based on data from the Baby Connectome Project demonstrate that our method outperforms state-of-the-art methods both qualitatively and quantitatively.

DeepDTI: High-fidelity six-direction diffusion tensor imaging using deep learning

Diffusion tensor magnetic resonance imaging (DTI) is unsurpassed in its ability to map tissue microstructure and structural connectivity in the living human brain. Nonetheless, the angular sampling requirement for DTI leads to long scan times and poses a critical barrier to performing high-quality DTI in routine clinical practice and large-scale research studies. In this work we present a new processing framework for DTI entitled DeepDTI that minimizes the data requirement of DTI to six diffusion-weighted images (DWIs) required by conventional voxel-wise fitting methods for deriving the six unique unknowns in a diffusion tensor using data-driven supervised deep learning. DeepDTI maps the input non-diffusion-weighted (b ​= ​0) image and six DWI volumes sampled along optimized diffusion-encoding directions, along with T1-weighted and T2-weighted image volumes, to the residuals between the input and high-quality output b = 0 image and DWI volumes using a 10-layer three-dimensional convolutional neural network (CNN). The inputs and outputs of DeepDTI are uniquely formulated, which not only enables residual learning to boost CNN performance but also enables tensor fitting of resultant high-quality DWIs to generate orientational DTI metrics for tractography. The very deep CNN used by DeepDTI leverages the redundancy in local and non-local spatial information and across diffusion-encoding directions and image contrasts in the data. The performance of DeepDTI was systematically quantified in terms of the quality of the output images, DTI metrics, DTI-based tractography and tract-specific analysis results. We demonstrate rotationally-invariant and robust estimation of DTI metrics from DeepDTI that are comparable to those obtained with two b ​= ​0 images and 21 DWIs for the primary eigenvector derived from DTI and two b ​= ​0 images and 26-30 DWIs for various scalar metrics derived from DTI, achieving 3.3-4.6 × ​acceleration, and twice as good as those of a state-of-the-art denoising algorithm at the group level. The twenty major white-matter tracts can be accurately identified from the tractography of DeepDTI results. The mean distance between the core of the major white-matter tracts identified from DeepDTI results and those from the ground-truth results using 18 ​b ​= ​0 images and 90 DWIs measures around 1-1.5 ​mm. DeepDTI leverages domain knowledge of diffusion MRI physics and power of deep learning to render DTI, DTI-based tractography, major white-matter tracts identification and tract-specific analysis more feasible for a wider range of neuroscientific and clinical studies.

Extracting diffusion tensor fractional anisotropy and mean diffusivity from 3-direction DWI scans using deep learning

PURPOSE

To develop and evaluate machine-learning methods that reconstruct fractional anisotropy (FA) values and mean diffusivities (MD) from 3-direction diffusion MRI (dMRI) acquisitions.

METHODS

Two machine-learning models were implemented to map undersampled dMRI signals with high-quality FA and MD maps that were reconstructed from fully sampled DTI scans. The first model was a previously described multilayer perceptron (MLP), which maps signals and FA/MD values from a single voxel. The second was a convolutional neural network U-Net model, which maps dMRI slices to full FA/MD maps. Each method was trained on dMRI brain scans (N = 46), and reconstruction accuracies were compared with conventional linear-least-squares (LLS) reconstructions.

RESULTS

In an independent testing cohort (N = 20), 3-direction U-Net reconstructions had significantly lower absolute FA error than both 3-direction MLP (U-Net_3-dir : 0.06 ± 0.01 vs. MLP_3-dir : 0.08 ± 0.01, P < 1 × 10^-5 ) and 6-direction LLS (LLS_6-dir : 0.09 ± 0.03, P = 1 × 10^-5 ). The MD errors were not significantly different among 3-direction MLP (0.06 ± 0.01 × 10^-3 mm² /s), 3-direction U-Net (0.06 ± 0.01 × 10^-3 mm² /s), and 6-direction LLS (0.07 ± 0.02 × 10^-3 mm² /s, P > .1).

CONCLUSION

The proposed U-Net model reconstructed FA from 3-direction dMRI scans with improved accuracy compared with both a previously described MLP approach and LLS fitting from 6-direction scans. The MD reconstruction accuracies did not differ significantly between reconstructions.