MR Acquisition Research Articles

A Deep Framework Assembling Principled Modules for CS-MRI: Unrolling Perspective, Convergence Behaviors, and Practical Modeling.

Compressed Sensing Magnetic Resonance Imaging (CS-MRI) significantly accelerates MR acquisition at a sampling rate much lower than the Nyquist criterion. A major challenge for CS-MRI lies in solving the severely ill-posed inverse problem to reconstruct aliasing-free MR images from the sparse k -space data. Conventional methods typically optimize an energy function, producing restoration of high quality, but their iterative numerical solvers unavoidably bring extremely large time consumption. Recent deep techniques provide fast restoration by either learning direct prediction to final reconstruction or plugging learned modules into the energy optimizer. Nevertheless, these data-driven predictors cannot guarantee the reconstruction following principled constraints underlying the domain knowledge so that the reliability of their reconstruction process is questionable. In this paper, we propose a deep framework assembling principled modules for CS-MRI that fuses learning strategy with the iterative solver of a conventional reconstruction energy. This framework embeds an optimal condition checking mechanism, fostering efficient and reliable reconstruction. We also apply the framework to three practical tasks, i.e., complex-valued data reconstruction, parallel imaging and reconstruction with Rician noise. Extensive experiments on both benchmark and manufacturer-testing images demonstrate that the proposed method reliably converges to the optimal solution more efficiently and accurately than the state-of-the-art in various scenarios.

Impact of different scanners and acquisition parameters on robustness of MR radiomics features based on women\u2019s cervix

MR Radiomics based on cervical lesions from one single scanner has achieved promising results. However, it is a challenge to achieve clinical translation. Considering multi-scanners and non-uniform scanning parameters from different centers in a real-world medical scenario, we should first identify the influence of such conditions on the robustness of MR radiomics features (RFs) based on the female cervix. In this study, 9 healthy female volunteers were enrolled and 3 kiwis were selected as references. Each of them underwent T2 weighted imaging in three different 3.0-T MR scanners with uniform acquisition parameters, and in one MR scanner with various scanning parameters. A total of 396 RFs were extracted from their images with and without decile intensity normalization. The RFs’ reproducibility was evaluated by coefficient of variation (CV) and quartile coefficient of dispersion (QCD). Representative features were selected using the hierarchical cluster analysis and their discrimination abilities were estimated by ROC analysis through retrospective comparison with the junctional zone and the outer muscular layer of healthy cervix in patients (n = 58) with leiomyoma. This study showed that only a few RFs were robust across different MR scanners and acquisition parameters based on females’ cervix, which might be improved by decile intensity normalization method.

Pilot tone-based motion correction for prospective respiratory compensated cardiac cine MRI.

To evaluate prospective motion correction using the pilot tone (PT) as a quantitative respiratory motion signal with high temporal resolution for cardiac cine images during free breathing. Before cine data acquisition, a short prescan was performed, calibrating the PT to the respiratory-induced heart motion using respiratory-resolved real-time images. The calibrated PT was then applied for nearly real-time prospective motion correction of cine MRI through slice tracking (ie, updating the slice position before every readout). Additionally, in-plane motion correction was performed retrospectively also based on the calibrated PT data. The proposed method was evaluated in a moving phantom and 10 healthy volunteers. The PT showed very good correlation to the phantom motion. In volunteer studies using a long-term scan over 7.96 ± 1.40 min, the mean absolute error between registered and predicted motion from the PT was 1.44 ± 0.46 mm in head-feet and 0.46 ± 0.07 mm in anterior-posterior direction. Irregular breathing could also be corrected well with the PT. The PT motion correction leads to a significant improvement of contrast-to-noise ratio by 68% (P ≤ .01) between blood pool and myocardium and sharpness of endocardium by 24% (P = .04) in comparison to uncorrected data. The image score, which refers to the cine image quality, has improved with the utilization of the proposed PT motion correction. The proposed approach provides respiratory motion-corrected cine images of the heart with improved image quality and a high scan efficiency using the PT. The PT is independent of the MR acquisition, making this a very flexible motion-correction approach.

Evaluation of an Early Regression Index (ERITCP) as Predictor of Pathological Complete Response in Cervical Cancer: A Pilot-Study

Background: Recent studies have highlighted the potentialities of a radiobiological parameter, the early regression index (ERITCP), in the treatment response prediction for rectal cancer patients treated with chemoradiotherapy followed by surgery. The aim of this study is to evaluate the performance of this parameter in predicting pathological complete response (pCR) in the context of low field MR guided radiotherapy (MRgRT) for cervical cancer (CC). Methods: A total of 16 patients affected by CC were enrolled. All patients underwent a MRgRT treatment, with prescription of 50.6 Gy in 22 fractions. A daily MR acquisition was performed at simulation and on each treatment fraction. Gross tumor volume (GTV) was delineated on the MR images acquired at the following biological effective dose (BED) levels: 14, 28, 42, 54 and 62 Gy. The ERITCP was calculated at the different BED levels and its predictive performance was quantified in terms of receiver operating characteristic (ROC) curve. Results: pCR was observed in 11/16 cases. The highest discriminative power of ERITCP was reported when a BED value of 28 Gy is reached, obtaining an area under curve (AUC) of 0.84. Conclusion: This study confirmed ERITCP as a promising response biomarker also for CC, although further studies with larger cohort of patients are recommended.

MRI-based radiomics to predict lipomatous soft tissue tumors malignancy: a pilot study

ObjectivesTo develop and validate a MRI-based radiomic method to predict malignancies in lipomatous soft tissue tumors.MethodsThis retrospective study searched in the database of our pathology department, data from patients with lipomatous soft tissue tumors, with histology and gadolinium-contrast enhanced T1w MR images, obtained from 56 centers with non-uniform protocols. For each tumor, 87 radiomic features were extracted by two independent observers to evaluate the inter-observer reproducibility. A reduction of learning base dimension was performed from reproducibility and relevancy criteria. A model was subsequently prototyped using a linear support vector machine to predict malignant lesions.ResultsEighty-one subjects with lipomatous soft tissue tumors including 40 lipomas and 41 atypical lipomatous tumors or well-differentiated liposarcomas with fat-suppressed T1w contrast enhanced MR images available were retrospectively enrolled. Based on a Pearson’s correlation coefficient threshold at 0.8, 55 out of 87 (63.2%) radiomic features were considered reproducible. Further introduction of relevancy finally selected 35 radiomic features to be integrated in the model. To predict malignant tumors, model diagnostic performances were as follow: AUROC = 0.96; sensitivity = 100%; specificity = 90%; positive predictive value = 90.9%; negative predictive value = 100% and overall accuracy = 95.0%.ConclusionThis work demonstrates that radiomics allows to predict malignancy in soft tissue lipomatous tumors with routinely used MR acquisition in clinical oncology. These encouraging results need to be further confirmed in an external validation population.

In vivo estimation of normal left ventricular stiffness and contractility based on routine cine MR acquisition

Post-myocardial infarction remodeling process is known to alter the mechanical properties of the heart. Biomechanical parameters, such as tissue stiffness and contractility, would be useful for clinicians to better assess the severity of the diseased heart. However, these parameters are difficult to obtain in the current clinical practice. In this paper, we estimated subject-specific in vivo myocardial stiffness and contractility from 21 healthy volunteers, based on left ventricle models constructed from data acquired from routine cardiac MR acquisition only. The subject-specific biomechanical parameters were quantified using an inverse finite-element modelling approach. The personalized models were evaluated against relevant clinical metrics extracted from the MR data, such as circumferential strain, wall thickness and fractional thickening. We obtained the ranges of healthy biomechanical indices of 1.60 ± 0.22 kPa for left ventricular stiffness and 95.13 ± 14.56 kPa for left ventricular contractility. These reference normal values can be used for future model-based investigation on the stiffness and contractility of ischemic myocardium.

7T versus 3T MR Angiography to Assess Unruptured Intracranial Aneurysms.

ABSTRACTBACKGROUND AND PURPOSEAneurysm size and neck measurements are important for treatment decisions. The introduction of 7T magnetic resonance angiography (MRA) led to new possibilities assessing aneurysm morphology and flow due to the higher signal‐to‐noise ratio. However, it is unknown if the size measurements on 7T MRA are similar to those on the standard 3T MRA. This study aimed to compare aneurysm size measurements between 7T and 3T MRA.METHODSWe included 18 patients with 22 aneurysms who underwent both 3T and 7T MRA. Three acquisition protocols were compared: 3T time of flight (TOF), 7T TOF, and 7T contrast‐enhanced MRA. Each aneurysm on each protocol was measured by at least two experienced neuroradiologists. Subsequently, the differences were evaluated using scatterplots and the intraclass correlation coefficients (ICC) of agreement.RESULTSThere was a good agreement among the neuroradiologists for the height and width measurements (mean ICC: .78‐.93); the neck measurements showed a moderate agreement with a mean ICC of .57‐.72. Between the MR acquisition protocols, there was a high agreement for all measurements with a mean ICC of .81‐.96. Measurement differences between acquisition protocols (0‐2.9 mm) were in the range of the differences between the neuroradiologists (0‐3.6 mm).CONCLUSIONOur study showed that 7T MRA, both nonenhanced and contrast‐enhanced, has a high agreement in aneurysm size measurements compared to 3T. This suggests that 7T is useful for reliable aneurysm size assessment.

Deep learning-based image reconstruction and motion estimation from undersampled radial k-space for real-time MRI-guided radiotherapy

To enable magnetic resonance imaging (MRI)-guided radiotherapy with real-time adaptation, motion must be quickly estimated with low latency. The motion estimate is used to adapt the radiation beam to the current anatomy, yielding a more conformal dose distribution. As the MR acquisition is the largest component of latency, deep learning (DL) may reduce the total latency by enabling much higher undersampling factors compared to conventional reconstruction and motion estimation methods. The benefit of DL on image reconstruction and motion estimation was investigated for obtaining accurate deformation vector fields (DVFs) with high temporal resolution and minimal latency.2D cine MRI acquired at 1.5 T from 135 abdominal cancer patients were retrospectively included in this study. Undersampled radial golden angle acquisitions were retrospectively simulated. DVFs were computed using different combinations of conventional- and DL-based methods for image reconstruction and motion estimation, allowing a comparison of four approaches to achieve real-time motion estimation. The four approaches were evaluated based on the end-point-error and root-mean-square error compared to a ground-truth optical flow estimate on fully-sampled images, the structural similarity (SSIM) after registration and time necessary to acquire k-space, reconstruct an image and estimate motion.The lowest DVF error and highest SSIM were obtained using conventional methods up to . For undersampling factors , the lowest DVF error and highest SSIM were obtained using conventional image reconstruction and DL-based motion estimation. We have found that, with this combination, accurate DVFs can be obtained up to with an average root-mean-square error up to 1 millimeter and an SSIM greater than 0.8 after registration, taking 60 milliseconds.High-quality 2D DVFs from highly undersampled k-space can be obtained with a high temporal resolution with conventional image reconstruction and a deep learning-based motion estimation approach for real-time adaptive MRI-guided radiotherapy.

The YOUth cohort study: MRI protocol and test-retest reliability in adults

The YOUth cohort study is a unique longitudinal study on brain development in the general population. As part of the YOUth study, 2000 children will be included at 8, 9 or 10 years of age and planned to return every three years during adolescence. Magnetic resonance imaging (MRI) brain scans are collected, including structural T1-weighted imaging, diffusion-weighted imaging (DWI), resting-state functional MRI and task-based functional MRI. Here, we provide a comprehensive report of the MR acquisition in YOUth Child & Adolescent including the test-retest reliability of brain measures derived from each type of scan. To measure test-retest reliability, 17 adults were scanned twice with a week between sessions using the full YOUth MRI protocol. Intraclass correlation coefficients were calculated to quantify reliability. Global brain measures derived from structural T1-weighted and DWI scans were reliable. Resting-state functional connectivity was moderately reliable, as well as functional brain measures for both the inhibition task (stop versus go) and the emotion task (face versus house). Our results complement previous studies by presenting reliability results of regional brain measures collected with different MRI modalities. YOUth facilitates data sharing and aims for reliable and high-quality data. Here we show that using the state-of-the art YOUth MRI protocol brain measures can be estimated reliably.

Deep Learning-Based Detection and Correction of Cardiac MR Motion Artefacts During Reconstruction for High-Quality Segmentation

Segmenting anatomical structures in medical images has been successfully addressed with deep learning methods for a range of applications. However, this success is heavily dependent on the quality of the image that is being segmented. A commonly neglected point in the medical image analysis community is the vast amount of clinical images that have severe image artefacts due to organ motion, movement of the patient and/or image acquisition related issues. In this paper, we discuss the implications of image motion artefacts on cardiac MR segmentation and compare a variety of approaches for jointly correcting for artefacts and segmenting the cardiac cavity. The method is based on our recently developed joint artefact detection and reconstruction method, which reconstructs high quality MR images from k-space using a joint loss function and essentially converts the artefact correction task to an under-sampled image reconstruction task by enforcing a data consistency term. In this paper, we propose to use a segmentation network coupled with this in an end-to-end framework. Our training optimises three different tasks: 1) image artefact detection, 2) artefact correction and 3) image segmentation. We train the reconstruction network to automatically correct for motion-related artefacts using synthetically corrupted cardiac MR k-space data and uncorrected reconstructed images. Using a test set of 500 2D+time cine MR acquisitions from the UK Biobank data set, we achieve demonstrably good image quality and high segmentation accuracy in the presence of synthetic motion artefacts. We showcase better performance compared to various image correction architectures.

Dynamic Fast Imaging Employing Steady State Acquisition Magnetic Resonance Imaging of the Vocal Tract in One Overtone Male Singer: Our Preliminary Experience

To demonstrate physiological changing of vocal tract's structures during overtone singing with commercial magnetic resonance imaging (MRI) Fast Imaging Employing Steady State Acquisition (FIESTA) dynamic sequence. A 1.5 T MRI with a 16 channel head-and-neck coil and a FIESTA sequence were used. A temporal resolution of 0.155 sec (7 image/s). A single professional singer was studied. The MR acquisition is made while the singer performed a predetermined singing sequence. Three different overtone singing techniques were examined (L-technique, J-technique, and NG technique) and one effect (Ezengileer) applied to L-technique. For each overtone technique we evaluated MRI movement of lips, tongue, velopharyngeal closure, and relationship among tongue and pharyngeal posterior wall/soft palate. To cancel the noise over-imposed, the dynamic MRI was subsequently dubbed in studio with the audio of the preset overtone sing. Dubbed MR images were analyzed with an Overtone Analyzer Software and different sound frequencies were identified and pointed out as colored lines. This study shows that different overtone techniques are related to a specific conformation of tongue, lips, soft and hard palate and motion's relation changing between them. Only a correct conformation of vocal tract's structure allows resonance and so to hear desired fundamental and harmonic pitch in overtone singing. The preliminary data of our study demonstrates that FIESTA dynamic MRI sequence can be used to depict changing of position of vocal tract's structure in overtone singing techniques with a good temporal and anatomic resolution.

M140. WHAT HAPPENS IN THE BRAIN A FEW SECONDS BEFORE THE ONSET AND OFFSET OF AN HALLUCINATORY EPISODE?

BackgroundAuditory verbal hallucinations (AVH), the sense of “hearing voices” in the absence of a corresponding auditory source, are a major symptom in schizophrenia. Hemodynamic and electrophysiology studies have shown aberrant activation in fronto-temporal speech and language areas (Curcic-Blake et al., 2017 for review) – indicating neurobiological mediators relating to onset and duration of AVH episodes. However, unanswered is the question of specific neuronal precursors to the spontaneous cessation of an AVH episode. We therefore studied what happens in the brain a few seconds before both the onset and offset of an AVH episode.MethodsStructural and functional MRI data were collected from 44 patients diagnosed with a schizophrenia spectrum disorder, at three sites (University of Bergen, Norway, n=12; Medical University of Plovdiv, Bulgaria, n=13; Utrecht University Medical Center, Netherlands, n=19, total 25m, 19f, age 37.9+/-13.2 years). Symptom severity was assessed with the PANSS; mean total (+/-SD) 64.9+/-16.9; P3 hallucinatory behaviour 4.6+/-1.1. Additional assessments (not reported), scanner hardware and particulars of the MR acquisition varied between sites.Functional data were collected with a “symptom-capture” paradigm (Sommer et al., 2008), where subjects were instructed to press a button when a hallucinatory episode began, and another when the episode ended. Functional data were processed using an FSL FEAT pipeline, with additional filtering using the ICA_AROMA method. Data were then subjected to statistical analysis, contrasting hallucinatory with non-hallucinatory periods; higher-level analysis allowed comparison with findings from previous meta-analyses (Kompus et al. 2011, Jardri et al., 2011).A number of regions of interest were selected based on an initial PCA, in addition to regions nominated in the meta-analyses. For each region, for each start- and end-of-hallucination event, a time-course was extracted for the interval t=-10 to +15sec with respect to button-press onset. Time-courses were aligned and group-averaged (rejecting outliers iteratively) to yield a model time-course for start- and end- hallucinatory events in each region. Permutation analysis (n=5000) was performed to identify differential effects between start and end events.ResultsFunctional block-analysis revealed several clusters of activation, including among others the left fronto-temporal language areas (superior temporal gyrus and Broca’s area). Similar patterns were present on a per-site basis.Time-course analysis demonstrated a differential effect in the left paracingulate sulcus, characterized by a significant dip in activation (Δ=-158iu, p=0.021) at the end-of-hallucination events, in contrast to a significant peak (Δ=35iu, p=0.0142) at the start-of-hallucination events. This activity preceded the motor response for the same events.DiscussionClusters identified in the block-analysis at group-level were consistent with findings from meta-analyses; this attests to the validity of the source data and initial processing. The novel finding of anticipatory neuronal activity in the paracingulate region, differentiated between onset and offset of AVH- events, complements recent findings of reduced functional connectivity (Alonso-Solís et al., 2015) and morphological differences (Garrison et al., 2019) in the same region. This suggests that the paracingulate region may be crucial in both the initiation and the cessation of an AVH-episode and speaks to a kind of regulatory role for the region. This may in turn point to differential action of excitatory and inhibitory transmitters at a lower level of explanation.

Measurement of Tongue Tip Velocity from Real-Time MRI and Phase-Contrast Cine-MRI in Consonant Production.

We evaluate velocity of the tongue tip with magnetic resonance imaging (MRI) using two independent approaches. The first one consists in acquisition with a real-time technique in the mid-sagittal plane. Tracking of the tongue tip manually and with a computer vision method allows its trajectory to be found and the velocity to be calculated as the derivative of the coordinate. We also propose to use another approach—phase contrast MRI—which enables velocities of the moving tissues to be measured directly. We recorded the sound simultaneously with the MR acquisition which enabled us to make conclusions regarding the relation between the movements and the sound. We acquired the data from two French-speaking subjects articulating /tata/. The results of both methods are in qualitative agreement and are consistent with other reviewer techniques used for evaluation of the tongue tip velocity.

MR Intracranial Vessel Wall Imaging: A Systematic Review.

The purpose of this systematic review is to identify trends and extent of variability in intracranial vessel wall MR imaging (VWI) techniques and protocols. Although variability in selection of protocol design and pulse sequence type is known, data on what and how protocols vary are unknown. Three databases were searched to identify publications using intracranial VWI. Publications were screened by predetermined inclusion/exclusion criteria. Technical development publications were scored for completeness of reporting using a modified Nature Reporting Summary Guideline to assess reproducibility. From 2,431 articles, 122 met the inclusion criteria. Trends over the last 23 years (1995-2018) show increased use of 3-Tesla MR (P<.001) and 3D volumetric T1-weighted acquisitions (P<.001). Most (65%) clinical VWI publications report achieving a noninterpolated in-plane spatial resolution of ≤.55 mm. In the last decade, an increasing number of technical development (n = 20) and 7 Tesla (n = 12) publications have been published, focused on pulse sequence development, improving cerebrospinal fluid suppression, scan efficiency, and imaging ex vivo specimen for histologic validation. Mean Reporting Summary Score for the technical development publications was high (.87, range: .63-1.0) indicating strong scientific technical reproducibility. Innovative work continues to emerge to address implementation challenges. Gradual adoption into the research and scientific community was suggested by a shift in the name in the literature from "high-resolution MR" to "vessel wall imaging," specifying diagnostic intent. Insight into current practices and identifying the extent of technical variability in the literature will help to direct future clinical and technical efforts to address needs for implementation.

Believing is seeing: an fMRI study of thought-action fusion in healthy male adults.

Thought-action fusion (TAF) is a tendency of individuals to establish causal relations between their own thoughts and external reality. TAF can lead to maladaptive behaviors typically observed in obsessional thoughts. However, neural mechanisms underlying TAF are still unknown. In this study, 38 healthy men were informed that MR signals were able to detect thoughts of the word 'apple' and that this recognition could result in the administration of electrical shocks to a person outside the scanner. During MR acquisition, they were asked to suppress or not suppress the thought of 'apple' while sham electrical shocks were or were not administered to the other person. The main effect of the sham administration of electrical shock to another person was shown in the bilateral lingual gyri, fusiform gyri, and middle occipital cortices (FDR corrected p < 0.05). Also, fusiform gyrus, lingual gyrus, and middle occipital cortex activity correlated with scores of guilty feeling only when participants consciously tried to think of apple as less as possible. Our study demonstrates that visual association areas may play primary roles in TAF. The simple belief and visual imagery that one's thought may lead to someone's injury activated visual areas of the brain where, in turn, brain activity is associated with feelings of guilt.

Three-dimensional accelerated acquisition for hyperpolarized 13 C MR with blipped stack-of-spirals and conjugate-gradient SENSE.

To test a new parallel imaging strategy for acceleration of hyperpolarized 13 C MR acquisitions based on a 3D blipped stack-of-spirals trajectory and conjugate-gradient SENSE reconstruction with precalibrated sensitivities. The blipped stack-of-spirals trajectory was developed for an acceleration factor of 4, based on an undersampled stack-of-spirals with gradient blips during spiral readout. The trajectory was developed with volumetric coverage of a large FOV and with high spatial resolution. High temporal resolution was attained through spectral-spatial excitation and 4 excitations per volume. The blipped stack-of-spirals was evaluated in simulations and phantom experiments. Next, the method was evaluated for kidney and cardiac imaging in 2 separate healthy pigs. Simulation and phantom results showed successful acquisition and reconstruction, but also revealed reconstruction challenges for certain locations and for wide signal sources. For the kidney experiment, the accelerated acquisition showed high similarity to 2 separately acquired fully sampled data sets with matched spatial and temporal resolution, respectively. For the cardiac experiment, the accelerated acquisition proved able to map each metabolite in 3 dimensions within a single cardiac cycle. The proposed method demonstrated effective mapping of metabolism in both kidneys and the heart of healthy pigs. Limitations seen in phantom experiments, may be irrelevant for most clinical applications, but should be kept in mind as well as reconstruction challenges related to residual aliasing. All in all, we show that the blipped stack-of-spirals is a relevant parallel imaging method for hyperpolarized human imaging, facilitating better insights into metabolism compared with nonaccelerated acquisition.

A Transfer-Learning Approach for Accelerated MRI Using Deep Neural Networks.

Neural networks have received recent interest for reconstruction of undersampled MR acquisitions. Ideally, network performance should be optimized by drawing the training and testing data from the same domain. In practice, however, large datasets comprising hundreds of subjects scanned under a common protocol are rare. The goal of this study is to introduce a transfer-learning approach to address the problem of data scarcity in training deep networks for accelerated MRI. Neural networks were trained on thousands(upto 4 thousand) of samples from public datasets of either natural images or brain MR images. The networks were then fine-tuned using only tens of brain MR images in a distinct testing domain. Domain-transferred networks were compared to networks trained directly in the testing domain. Network performance was evaluated for varying acceleration factors (4-10), number of training samples (0.5-4k), and number of fine-tuning samples (0-100). The proposed approach achieves successful domain transfer between MR images acquired with different contrasts (T1 - and T2 -weighted images) and between natural and MR images (ImageNet and T1 - or T2 -weighted images). Networks obtained via transfer learning using only tens of images in the testing domain achieve nearly identical performance to networks trained directly in the testing domain using thousands(upto 4 thousand) of images. The proposed approach might facilitate the use of neural networks for MRI reconstruction without the need for collection of extensive imaging datasets.

The noise navigator: a surrogate for respiratory-correlated 4D-MRI for motion characterization in radiotherapy

Respiratory-correlated 4D-MRI can characterize respiratory-induced motion of tumors and organs-at-risk for radiotherapy treatment planning and is a necessity for image guidance of moving tumors treated on an MRI-linac. Essential for 4D-MRI generation is a robust respiratory surrogate signal.We investigated the feasibility of the noise navigator as respiratory surrogate signal for 4D-MRI generation. The noise navigator is based on the respiratory-induced modulation of the thermal noise variance measured by the receive coils during MR acquisition and thus is inherently present and synchronized with MRI data acquisition. Additionally, the noise navigator can be combined with any rectilinear readout strategy (e.g. radial and cartesian) and is independent of MR image contrast and imaging orientation. For radiotherapy applications, the noise navigator provides a robust respiratory signal for patients scanned with an elevated coil setup. This is particularly attractive for widely used cartesian sequences where currently a non-interfering self-navigation means is lacking for MRI-based simulation and MRI-guided radiotherapy.The feasibility of 4D-MRI generation with the noise navigator as respiratory surrogate signal was demonstrated for both cartesian and radial readout strategies in radiotherapy setup on four healthy volunteers and two radiotherapy patients on a dedicated 1.5 T MRI scanner and two radiotherapy patients on a 1.5 T MRI-linac system. Moreover, the respiratory-correlated 4D-MR images showed liver motion comparable to a reference 2D cine MRI series for the volunteers.For 2D cartesian cine MRI acquisitions, both the noise navigator and respiratory bellows were benchmarked against an image navigator. Respiratory phase detection based on the noise navigator agreed 1.4 times better with the image navigator than the respiratory bellows did. For a 3D Stack-of-Stars acquisitions, the noise navigator was compared to radial self-navigation and a 1.7 times higher respiratory phase detection agreement was observed than for the respiratory bellows compared to radial self-navigation.

Deep-Learning Methods for Parallel Magnetic Resonance Imaging Reconstruction: A Survey of the Current Approaches, Trends, and Issues.

Following the success of deep learning in a wide range of applications, neural network-based machine learning techniques have received interest as a means of accelerating magnetic resonance imaging (MRI). A number of ideas inspired by deep learning techniques from computer vision and image processing have been successfully applied to non-linear image reconstruction in the spirit of compressed sensing for both low dose computed tomography and accelerated MRI. The additional integration of multi-coil information to recover missing k-space lines in the MRI reconstruction process, is still studied less frequently, even though it is the de-facto standard for currently used accelerated MR acquisitions. This manuscript provides an overview of the recent machine learning approaches that have been proposed specifically for improving parallel imaging. A general background introduction to parallel MRI is given that is structured around the classical view of image space and k-space based methods. Both linear and non-linear methods are covered, followed by a discussion of recent efforts to further improve parallel imaging using machine learning, and specifically using artificial neural networks. Image-domain based techniques that introduce improved regularizers are covered as well as k-space based methods, where the focus is on better interpolation strategies using neural networks. Issues and open problems are discussed as well as recent efforts for producing open datasets and benchmarks for the community.

An Adaptive Intelligence Algorithm for Undersampled Knee MRI Reconstruction

Adaptive intelligence aims at empowering machine learning techniques with the additional use of domain knowledge. In this work, we present the application of adaptive intelligence to accelerate MR acquisition. Starting from undersampled k-space data, an iterative learning-based reconstruction scheme inspired by compressed sensing theory is used to reconstruct the images. We developed a novel deep neural network to refine and correct prior reconstruction assumptions given the training data. The network was trained and tested on a knee MRI dataset from the 2019 fastMRI challenge organized by Facebook AI Research and NYU Langone Health. All submissions to the challenge were initially ranked based on similarity with a known groundtruth, after which the top 4 submissions were evaluated radiologically. Our method was evaluated by the fastMRI organizers on an independent challenge dataset. It ranked #1, shared #1, and #3 on respectively the 8× accelerated multi-coil, the 4× multi-coil, and the 4× single-coil tracks. This demonstrates the superior performance and wide applicability of the method.