dtiRIM: A generalisable deep learning method for diffusion tensor imaging.

Diffusion weighted MRI is an indispensable tool for routine patient screening and diagnostics of pathology. Recently, several deep learning methods have been proposed to quantify diffusion parameters, but poor generalisation to new data prevents broader use of these methods, as they require retraining of the neural network for each new scan protocol. In this work, we present the dtiRIM, a new deep learning method for Diffusion Tensor Imaging (DTI) based on the Recurrent Inference Machines. Thanks to its ability to learn how to solve inverse problems and to use the diffusion tensor model to promote data consistency, the dtiRIM can generalise to variations in the acquisition settings. This enables a single trained network to produce high quality tensor estimates for a variety of cases. We performed extensive validation of our method using simulation and in vivo data, and compared it to the Iterated Weighted Linear Least Squares (IWLLS), the approach of the state-of-the-art MRTrix3 software, and to an implementation of the Maximum Likelihood Estimator (MLE). Our results show that dtiRIM predictions present low dependency on tissue properties, anatomy and scanning parameters, with results comparable to or better than both IWLLS and MLE. Further, we demonstrate that a single dtiRIM model can be used for a diversity of data sets without significant loss in quality, representing, to our knowledge, the first generalisable deep learning based solver for DTI.

Accuracy and repeatability of joint sparsity multi-component estimation in MR Fingerprinting.

MR fingerprinting (MRF) is a promising method for quantitative characterization of tissues. Often, voxel-wise measurements are made, assuming a single tissue-type per voxel. Alternatively, the Sparsity Promoting Iterative Joint Non-negative least squares Multi-Component MRF method (SPIJN-MRF) facilitates tissue parameter estimation for identified components as well as partial volume segmentations. The aim of this paper was to evaluate the accuracy and repeatability of the SPIJN-MRF parameter estimations and partial volume segmentations. This was done (1) through numerical simulations based on the BrainWeb phantoms and (2) using in vivo acquired MRF data from 5 subjects that were scanned on the same week-day for 8 consecutive weeks. The partial volume segmentations of the SPIJN-MRF method were compared to those obtained by two conventional methods: SPM12 and FSL. SPIJN-MRF showed higher accuracy in simulations in comparison to FSL- and SPM12-based segmentations: Fuzzy Tanimoto Coefficients (FTC) comparing these segmentations and Brainweb references were higher than 0.95 for SPIJN-MRF in all the tissues and between 0.6 and 0.7 for SPM12 and FSL in white and gray matter and between 0.5 and 0.6 in CSF. For the in vivo MRF data, the estimated relaxation times were in line with literature and minimal variation was observed. Furthermore, the coefficient of variation (CoV) for estimated tissue volumes with SPIJN-MRF were 10.5% for the myelin water, 6.0% for the white matter, 5.6% for the gray matter, 4.6% for the CSF and 1.1% for the total brain volume. CoVs for CSF and total brain volume measured on the scanned data for SPIJN-MRF were in line with those obtained with SPM12 and FSL. The CoVs for white and gray matter volumes were distinctively higher for SPIJN-MRF than those measured with SPM12 and FSL. In conclusion, the use of SPIJN-MRF provides accurate and precise tissue relaxation parameter estimations taking into account intrinsic partial volume effects. It facilitates obtaining tissue fraction maps of prevalent tissues including myelin water which can be relevant for evaluating diseases affecting the white matter.

Differences Between MR Brain Region Segmentation Methods: Impact on Single-Subject Analysis.

For the segmentation of magnetic resonance brain images into anatomical regions, numerous fully automated methods have been proposed and compared to reference segmentations obtained manually. However, systematic differences might exist between the resulting segmentations, depending on the segmentation method and underlying brain atlas. This potentially results in sensitivity differences to disease and can further complicate the comparison of individual patients to normative data. In this study, we aim to answer two research questions: 1) to what extent are methods interchangeable, as long as the same method is being used for computing normative volume distributions and patient-specific volumes? and 2) can different methods be used for computing normative volume distributions and assessing patient-specific volumes? To answer these questions, we compared volumes of six brain regions calculated by five state-of-the-art segmentation methods: Erasmus MC (EMC), FreeSurfer (FS), geodesic information flows (GIF), multi-atlas label propagation with expectation-maximization (MALP-EM), and model-based brain segmentation (MBS). We applied the methods on 988 non-demented (ND) subjects and computed the correlation (PCC-v) and absolute agreement (ICC-v) on the volumes. For most regions, the PCC-v was good ( > 0.75 ), indicating that volume differences between methods in ND subjects are mainly due to systematic differences. The ICC-v was generally lower, especially for the smaller regions, indicating that it is essential that the same method is used to generate normative and patient data. To evaluate the impact on single-subject analysis, we also applied the methods to 42 patients with Alzheimer's disease (AD). In the case where the normative distributions and the patient-specific volumes were calculated by the same method, the patient's distance to the normative distribution was assessed with the z-score. We determined the diagnostic value of this z-score, which showed to be consistent across methods. The absolute agreement on the AD patients' z-scores was high for regions of thalamus and putamen. This is encouraging as it indicates that the studied methods are interchangeable for these regions. For regions such as the hippocampus, amygdala, caudate nucleus and accumbens, and globus pallidus, not all method combinations showed a high ICC-z. Whether two methods are indeed interchangeable should be confirmed for the specific application and dataset of interest.

Recurrent inference machines as inverse problem solvers for MR relaxometry.

In this paper, we propose the use of Recurrent Inference Machines (RIMs) to perform T1 and T2 mapping. The RIM is a neural network framework that learns an iterative inference process based on the signal model, similar to conventional statistical methods for quantitative MRI (QMRI), such as the Maximum Likelihood Estimator (MLE). This framework combines the advantages of both data-driven and model-based methods, and, we hypothesize, is a promising tool for QMRI. Previously, RIMs were used to solve linear inverse reconstruction problems. Here, we show that they can also be used to optimize non-linear problems and estimate relaxometry maps with high precision and accuracy. The developed RIM framework is evaluated in terms of accuracy and precision and compared to an MLE method and an implementation of the Residual Neural Network (ResNet). The results show that the RIM improves the quality of estimates compared to the other techniques in Monte Carlo experiments with simulated data, test-retest analysis of a system phantom, and in-vivo scans. Additionally, inference with the RIM is 150 times faster than the MLE, and robustness to (slight) variations of scanning parameters is demonstrated. Hence, the RIM is a promising and flexible method for QMRI. Coupled with an open-source training data generation tool, it presents a compelling alternative to previous methods.

The Correlation Between Wall Shear Stress and Plaque Composition in Advanced Human Carotid Atherosclerosis.

The role of wall shear stress (WSS) in atherosclerotic plaque development is evident, but the relation between WSS and plaque composition in advanced atherosclerosis, potentially resulting in plaque destabilization, is a topic of discussion. Using our previously developed image registration pipeline, we investigated the relation between two WSS metrics, time-averaged WSS (TAWSS) and the oscillatory shear index (OSI), and the local histologically determined plaque composition in a set of advanced human carotid plaques. Our dataset of 11 carotid endarterectomy samples yielded 87 histological cross-sections, which yielded 511 radial bins for analysis. Both TAWSS and OSI values were subdivided into patient-specific low, mid, and high tertiles. This cross-sectional study shows that necrotic core (NC) size and macrophage area are significantly larger in areas exposed to high TAWSS or low OSI. Local TAWSS and OSI tertile values were generally inversely related, as described in the literature, but other combinations were also found. Investigating the relation between plaque vulnerability features and different combinations of TAWSS and OSI tertile values revealed a significantly larger cap thickness in areas exposed to both low TAWSS and low OSI. In conclusion, our study confirmed previous findings, correlating high TAWSS to larger macrophage areas and necrotic core sizes. In addition, our study demonstrated new relations, correlating low OSI to larger macrophage areas, and a combination of low TAWSS and low OSI to larger cap thickness.

Fractional order vs. exponential fitting in UTE MR imaging of the patellar tendon.

PURPOSE: Quantification of the T2∗ relaxation time constant is relevant in various magnetic resonance imaging applications. Mono- or bi-exponential models are typically used to determine these parameters. However, in case of complex, heterogeneous tissues these models could lead to inaccurate results. We compared a model, provided by the fractional-order extension of the Bloch equation with the conventional models. METHODS: Axial 3D ultra-short echo time (UTE) scans were acquired using a 3.0 T MRI and a 16-channel surface coil. After image registration, voxel-wise T2∗ was quantified with mono-exponential, bi-exponential and fractional-order fitting. We evaluated all three models repeatability and the bias of their derived parameters by fitting at various noise levels. To investigate the effect of the SNR for the different models, a Monte-Carlo experiment with 1000 repeats was performed for different noise levels for one subject. For a cross-sectional investigation, we used the mean fitted values of the ROIs in five volunteers. RESULTS: Comparing the mono-exponential and the fractional order T2∗ maps, the fractional order fitting method yielded enhanced contrast and an improved delineation of the different tissues. In the case of the bi-exponential method, the long T2∗ component map demonstrated the anatomy clearly with high contrast. Simulations showed a nonzero bias of the parameters for all three mathematical models. ROI based fitting showed that the T2∗ values were different depending on the applied method, and they differed most for the patellar tendon in all subjects. CONCLUSIONS: In high SNR cases, the fractional order and bi-exponential models are both performing well with low bias. However, in all observed cases, one of the bi-exponential components has high standard deviation in T2∗. The bi-exponential model is suitable for T2∗ mapping, but we recommend using the fractional order model for cases of low SNR.

Orientation Prior and Consistent Model Selection Increase Sensitivity of Tract-Based Spatial Statistics in Crossing-Fiber Regions.

The goal of this paper is to increase the statistical power of crossing-fiber statistics in voxelwise analyses of diffusion-weighted magnetic resonance imaging (DW-MRI) data. In the proposed framework, a fiber orientation atlas and a model complexity atlas were used to fit the ball-and-sticks model to diffusion-weighted images of subjects in a prospective population-based cohort study. Reproducibility and sensitivity of the partial volume fractions in the ball-and-sticks model were analyzed using TBSS (tract-based spatial statistics) and compared to a reference framework. The reproducibility was investigated on two scans of 30 subjects acquired with an interval of approximately three weeks by studying the intraclass correlation coefficient (ICC). The sensitivity to true biological effects was evaluated by studying the regression with age on 500 subjects from 65 to 90 years old. Compared to the reference framework, the ICC improved significantly when using the proposed framework. Higher t-statistics indicated that regression coefficients with age could be determined more precisely with the proposed framework and more voxels correlated significantly with age. The application of a fiber orientation atlas and a model complexity atlas can significantly improve the reproducibility and sensitivity of crossing-fiber statistics in TBSS.

An MRI-based method to register patient-specific wall shear stress data to histology.

Wall shear stress (WSS), the frictional force exerted on endothelial cells by blood flow, is hypothesised to influence atherosclerotic plaque growth and composition. We developed a methodology for image registration of MR and histology images of advanced human carotid plaques and corresponding WSS data, obtained by MRI and computational fluid dynamics. The image registration method requires four types of input images, in vivo MRI, ex vivo MRI, photographs of transversally sectioned plaque tissue and histology images. These images are transformed to a shared 3D image domain by applying a combination of rigid and non-rigid registration algorithms. Transformation matrices obtained from registration of these images are used to transform subject-specific WSS data to the shared 3D image domain as well. WSS values originating from the 3D WSS map are visualised in 2D on the corresponding lumen locations in the histological sections and divided into eight radial segments. In each radial segment, the correlation between WSS values and plaque composition based on histological parameters can be assessed. The registration method was successfully applied to two carotid endarterectomy specimens. The resulting matched contours from the imaging modalities had Hausdorff distances between 0.57 and 0.70 mm, which is in the order of magnitude of the in vivo MRI resolution. We simulated the effect of a mismatch in the rigid registration of imaging modalities on WSS results by relocating the WSS data with respect to the stack of histology images. A 0.6 mm relocation altered the mean WSS values projected on radial bins on average by 0.59 Pa, compared to the output of original registration. This mismatch of one image slice did not change the correlation between WSS and plaque thickness. In conclusion, we created a method to investigate correlations between WSS and plaque composition.

Compressed Sensing 3D-GRASE for faster High-Resolution MRI.

PURPOSE: High-resolution three-dimensional (3D) structural MRI is useful for delineating complex or small structures of the body. However, it requires long acquisition times and high SAR, limiting its clinical use. The purpose of this work is to accelerate the acquisition of high-resolution images by combining compressed sensing and parallel imaging (CSPI) on a 3D-GRASE sequence and to compare it with a (CS)PI 3D-FSE sequence. Several sampling patterns were investigated to assess their influence on image quality. METHODS: The proposed k-space sampling patterns are based on two undersampled k-space grids, variable density (VD) Poisson-disc, and VD pseudo-random Gaussian, and five different trajectories described in the literature. Bloch simulations are performed to obtain the transform point spread function and evaluate the coherence of each sampling pattern. Image resolution was assessed by the full-width at half-maximum (FWHM). Prospective CSPI 3D-GRASE phantom and in vivo experiments in knee and brain are carried out to assess image quality, SNR, SAR, and acquisition time compared to PI 3D-GRASE, PI 3D-FSE, and CSPI 3D-FSE acquisitions. RESULTS: Sampling patterns with VD Poisson-disc obtain the lowest coherence for both PD-weighted and T2 -weighted acquisitions. VD pseudo-random Gaussian obtains lower FWHM, but higher sidelobes than VD Poisson-disc. CSPI 3D-GRASE reduces acquisition time (43% for PD-weighted and 40% for T2 -weighted) and SAR (∼45% for PD-weighted and T2 -weighted) compared to CSPI 3D-FSE. CONCLUSIONS: CSPI 3D-GRASE reduces acquisition time compared to a CSPI 3DFSE acquisition, preserving image quality. The design of the sampling pattern is crucial for image quality in CSPI 3D-GRASE image acquisitions.

A spatio-temporal reference model of the aging brain.

Both normal aging and neurodegenerative disorders such as Alzheimer's disease (AD) cause morphological changes of the brain. It is generally difficult to distinguish these two causes of morphological change by visual inspection of magnetic resonance (MR) images. To facilitate making this distinction and thus aid the diagnosis of neurodegenerative disorders, we propose a method for developing a spatio-temporal model of morphological differences in the brain due to normal aging. The method utilizes groupwise image registration to characterize morphological variation across brain scans of people with different ages. To extract the deformations that are due to normal aging we use partial least squares regression, which yields modes of deformations highly correlated with age, and corresponding scores for each input subject. Subsequently, we determine a distribution of morphologies as a function of age by fitting smooth percentile curves to these scores. This distribution is used as a reference to which a person's morphology score can be compared. We validate our method on two different datasets, using images from both cognitively normal subjects and patients with Alzheimer disease (AD). Results show that the proposed framework extracts the expected atrophy patterns. Moreover, the morphology scores of cognitively normal subjects are on average lower than the scores of AD subjects, indicating that morphology differences between AD subjects and healthy subjects can be partly explained by accelerated aging. With our methods we are able to assess accelerated brain aging on both population and individual level. A spatio-temporal aging brain model derived from 988 T1-weighted MR brain scans from a large population imaging study (age range 45.9-91.7y, mean age 68.3y) is made publicly available at www.agingbrain.nl.

Value of quantitative MRI parameters in predicting and evaluating clinical outcome in conservatively treated patients with chronic midportion Achilles tendinopathy: A prospective study.

OBJECTIVES: To evaluate whether baseline MRI parameters provide prognostic value for clinical outcome, and to study correlation between MRI parameters and clinical outcome. DESIGN: Observational prospective cohort study. METHODS: Patients with chronic midportion Achilles tendinopathy were included and performed a 16-week eccentric calf-muscle exercise program. Outcome measurements were the validated Victorian Institute of Sports Assessment-Achilles (VISA-A) questionnaire and MRI parameters at baseline and after 24 weeks. The following MRI parameters were assessed: tendon volume (Volume), tendon maximum cross-sectional area (CSA), tendon maximum anterior-posterior diameter (AP), and signal intensity (SI). Intra-class correlation coefficients (ICCs) and minimum detectable changes (MDCs) for each parameter were established in a reliability analysis. RESULTS: Twenty-five patients were included and complete follow-up was achieved in 20 patients. The average VISA-A scores increased significantly with 12.3 points (27.6%). The reliability was fair-good for all MRI-parameters with ICCs>0.50. Average tendon volume and CSA decreased significantly with 0.28cm3 (5.2%) and 4.52mm2 (4.6%) respectively. Other MRI parameters did not change significantly. None of the baseline MRI parameters were univariately associated with VISA-A change after 24 weeks. MRI SI increase over 24 weeks was positively correlated with the VISA-A score improvement (B=0.7, R2=0.490, p=0.02). CONCLUSIONS: Tendon volume and CSA decreased significantly after 24 weeks of conservative treatment. As these differences were within the MDC limits, they could be a result of a measurement error. Furthermore, MRI parameters at baseline did not predict the change in symptoms, and therefore have no added value in providing a prognosis in daily clinical practice.

PCA-based groupwise image registration for quantitative MRI.

Quantitative magnetic resonance imaging (qMRI) is a technique for estimating quantitative tissue properties, such as the T1 and T2 relaxation times, apparent diffusion coefficient (ADC), and various perfusion measures. This estimation is achieved by acquiring multiple images with different acquisition parameters (or at multiple time points after injection of a contrast agent) and by fitting a qMRI signal model to the image intensities. Image registration is often necessary to compensate for misalignments due to subject motion and/or geometric distortions caused by the acquisition. However, large differences in image appearance make accurate image registration challenging. In this work, we propose a groupwise image registration method for compensating misalignment in qMRI. The groupwise formulation of the method eliminates the requirement of choosing a reference image, thus avoiding a registration bias. The method minimizes a cost function that is based on principal component analysis (PCA), exploiting the fact that intensity changes in qMRI can be described by a low-dimensional signal model, but not requiring knowledge on the specific acquisition model. The method was evaluated on 4D CT data of the lungs, and both real and synthetic images of five different qMRI applications: T1 mapping in a porcine heart, combined T1 and T2 mapping in carotid arteries, ADC mapping in the abdomen, diffusion tensor mapping in the brain, and dynamic contrast-enhanced mapping in the abdomen. Each application is based on a different acquisition model. The method is compared to a mutual information-based pairwise registration method and four other state-of-the-art groupwise registration methods. Registration accuracy is evaluated in terms of the precision of the estimated qMRI parameters, overlap of segmented structures, distance between corresponding landmarks, and smoothness of the deformation. In all qMRI applications the proposed method performed better than or equally well as competing methods, while avoiding the need to choose a reference image. It is also shown that the results of the conventional pairwise approach do depend on the choice of this reference image. We therefore conclude that our groupwise registration method with a similarity measure based on PCA is the preferred technique for compensating misalignments in qMRI.

General and efficient super-resolution method for multi-slice MRI.

In this paper, a method is developed that reconstructs a high resolution image from an arbitrary set of multi-slice 3D MR images with a high in-plane resolution and a low through-plane resolution. Such images are often recorded to increase the efficiency of the acquisition. With a model of the acquisition of MR images, which is improved compared to previous super-resolution methods for MR images, a large system with linear equations is obtained. With the conjugated gradient method and this linear system, a high resolution image is reconstructed from MR images of an object. Also, a new and efficient method to apply an affine transformation to multi-dimensional images is presented. This method is used to efficiently reconstruction the high resolution image from multislice MR images with arbitrary orientations of the slices.

Likelihood-based hypothesis tests for brain activation detection from MRI data disturbed by colored noise: a simulation study.

Functional magnetic resonance imaging (fMRI) data that are corrupted by temporally colored noise are generally preprocessed (i.e., prewhitened or precolored) prior to functional activation detection. In this paper, we propose likelihood-based hypothesis tests that account for colored noise directly within the framework of functional activation detection. Three likelihood-based tests are proposed: the generalized likelihood ratio (GLR) test, the Wald test, and the Rao test. The fMRI time series is modeled as a linear regression model, where one regressor describes the task-related hemodynamic response, one regressor accounts for a constant baseline and one regressor describes potential drift. The temporal correlation structure of the noise is modeled as an autoregressive (AR) model. The order of the AR model is determined from practical null data sets using Akaike's information criterion (with penalty factor 3) as order selection criterion. The tests proposed are based on exact expressions for the likelihood function of the data. Using Monte Carlo simulation experiments, the performance of the proposed tests is evaluated in terms of detection rate and false alarm rate properties and compared to the current general linear model (GLM) test, which estimates the coloring of the noise in a separate step. Results show that theoretical asymptotic distributions of the GLM, GLR, and Wald test statistics cannot be reliably used for computing thresholds for activation detection from finite length time series. Furthermore, it is shown that, for a fixed false alarm rate, the detection rate of the proposed GLR test statistic is slightly, but (statistically) significantly improved compared to that of the common GLM-based tests. Finally, simulations results reveal that all tests considered show seriously inferior performance if the order of the AR model is not chosen sufficiently high to give an adequate description of the correlation structure of the noise, whereas the effects of (slightly) overmodeling are observed to be less harmful.