Search

In this work, we address the problem of creating a 3D dynamic atlas of the vocal tract that captures the dynamics of the articulators in all three dimensions in order to create a global speaker model independent of speaker-specific characteristics. The core steps of the proposed method are the temporal alignment of the real-time MR images acquired in several sagittal planes and their combination with adaptive kernel regression. As a preprocessing step, a reference space was created to be used in order to remove anatomical information of the speakers and keep only the variability in speech production for the construction of the atlas. The adaptive kernel regression makes the choice of atlas time points independently of the time points of the frames that are used as an input for the construction. The evaluation of this atlas construction method was made by mapping two new speakers to the atlas and by checking how similar the resulting mapped images are. The use of the atlas helps in reducing subject variability. The results show that the use of the proposed atlas can capture the dynamic behavior of the articulators and is able to generalize the speech production process by creating a universal-speaker reference space.

International audience; installation which might be impossible in some cases (e.g. pregnant women in lateral position, utilization of a tight flexible coil), and allows monitoring of only respiratory and heart activities. Moreover, being designed for clinical use, they offer no control over data acquisition and post-processing, which can be critical for some applications. For example, standard ECG processing can lead to incorrect QRS detection in case of steep gradient waveforms; no control over respiratory belt signal gain and offset (to avoid signal saturation) is available. Thus, often, accessory devices are thirdparty systems, which offer more flexibility, however have no communication interface with the MR system, or limited ones. Dedicated computer and electronics systems, as well as communication interfaces, are needed for use in either realtime/online applications or retrospective/offline applications. In the following we review several exemplary applications with the aim to derive typical requirements for such systems. Example applications that require real-time processing and communication include ECG denoising (7-10) and detection of the QRS complex (11-13), in order to trigger the MR sequence to the heart rate prospectively. The trigger is provided to the MR system by a TTL signal. The delays introduced by the real-time processing system, as defined by the latency (mean delay) and jitter (standard deviation of the delay), should be significantly lower that the temporal resolution of cardiac MR imaging, which is typically above 20 ms. Real-time display of the denoised ECG is also desired during anesthesia, cardiac stress MRI, or interventional MRI. Prospective motion correction (14) using external devices also relies on a real-time setup to modulate the MRI pulses (RF and magnetic field gradient pulses) in order to compensate for the effect of patient motion during spatial encoding. Low latency is also desired, ideally to be equal to the repetition time (TR) of the sequence or lower, i.e. 3 to 5 ms, though in practice it is limited by the time needed to process video images, e.g. 60 to 150 ms in (15). Software communication interfaces need to be defined with the vendor-dependent pulse sequence program in order to provide certain motion parameters (e.g. translations, rotations).

The study of articulatory gestures has a wide spectrum of applications, notably in speech production and recognition. Sets of phonemes, as well as their articulation, are language-specific; however, existing MRI databases mostly include English speakers. In our present work, we introduce a dataset acquired with MRI from 10 healthy native French speakers. A corpus consisting of synthetic sentences was used to ensure a good coverage of the French phonetic context. A real-time MRI technology with temporal resolution of 20 ms was used to acquire vocal tract images of the participants speaking. The sound was recorded simultaneously with MRI, denoised and temporally aligned with the images. The speech was transcribed to obtain phoneme-wise segmentation of sound. We also acquired static 3D MR images for a wide list of French phonemes. In addition, we include annotations of spontaneous swallowing., Measurement(s)Vocal tract images • SpeechTechnology Type(s)Magnetic Resonance Imaging • Microphone DeviceSample Characteristic - OrganismHomo sapiens Machine-accessible metadata file describing the reported data: 10.6084/m9.figshare.16404453

International audience; In this work, we address the prediction of speech articulators' temporal geometric position from the sequence of phonemes to be articulated. We start from a set of real-time MRI sequences uttered by a female French speaker. The contours of five articulators were tracked automatically in each of the frames in the MRI video. Then, we explore the capacity of a bidirectional GRU to correctly predict each articulator's shape and position given the sequence of phonemes and their duration. We propose a 5-fold cross-validation experiment to evaluate the generalization capacity of the model. In a second experiment, we evaluate our model's data efficiency by reducing training data. We evaluate the point-to-point Euclidean distance and the Pearson's correlations along time between the predicted and the target shapes. We also evaluate produced shapes of the critical articulators of specific phonemes. We show that our model can achieve good results with minimal data, producing very realistic vocal tract shapes.

International audience

International audience; In this paper we propose an algorithm for estimating vocal tract para sagittal slices in order to have a better overview of the behaviour of the articulators during speech production. The first step is to align the consonant-vowel (CV) data of the sagittal plains between them for the train speaker. Sets of transformations that connect the midsagittal frames with the neighbouring ones is acquired for the train speaker. Another set of transformations is calculated which transforms the midsagittal frames of the train speaker to the corresponding midsagittal frames of the test speaker and is used to adapt to the test speaker domain the previously computed sets of transformations. The newly adapted transformations are applied to the midsagittal frames of the test speaker in order to estimate the neighbouring sagittal frames. Several mono speaker models are combined to produce the final frame estimation. To evaluate the results, image cross-correlation between the original and the estimated frames was used. Results show good agreement between the original and the estimated frames.

International audience; A set of rtMR image transformations across time is computed during the production of CV that is afterwards applied to a new speaker in order to synthesize his/her CV pseudo rtMRI data. Synthesized images are compared with the original ones using image cross-correlation. 2 Purpose To be able to enlarge MRI speech corpus by synthesizing data.

International audience

International audience; Tongue contour extraction from real-time magnetic resonance images is a nontrivial task due to the presence of artifacts manifesting in form of blurring or ghostly contours. In this work, we present results of automatic tongue delineation achieved by means of U-Net auto-encoder convolutional neural network. We present both intra- and inter-subject validation. We used real-time magnetic resonance images and manually annotated 1-pixel wide contours as inputs. Predicted probability maps were post-processed in order to obtain 1-pixel wide tongue contours. The results are very good and slightly outperform published results on automatic tongue segmentation.

International audience; PurposeMultiple medical‐device leads implanted next to each other are often encountered in clinical practice. The aim of this work is to study a coupled transfer function model to evaluate the safety of these coupled leads submitted to the RF field of a 1.5T MRI scanner for a constant distance between both leads.MethodsThe effect of coupling on the heating of 2 cables with different termination conditions is evaluated experimentally. The coupled and single transfer functions are determined experimentally and used to predict the relative temperature increases of both cables alone and coupled. Two different coupled models, an additive model and a global model, are proposed. The coupled transfer functions are also simulated.ResultsThe coupling between cables has a strong influence on the resulting heating at the electrodes. The coupled additive transfer function model is a relevant tool to evaluate the heating of coupled leads separated by a constant distance. The global model underestimates the heating in one of the coupled cases by about 30%. The measured coupled transfer functions coincide with the simulated models.ConclusionIt is necessary to take into account the coupling effect between leads to evaluate the safety of implanted devices. This work shows that, in the case of 2 cables separated by a constant distance, that an experimentally determined coupled transfer function allows estimation of the heating of the 2 electrodes for a given incident field. Further work should take into account the in vivo varying distance between the 2 cables.

A Magnetic Resonance Imaging (MRI) scanner uses three different electromagnetic fields (EMF) to produce body images: a static permanent magnetic field (MF), several pulsed magnetic gradients, and a radiofrequency pulse. As a result, any occupation that includes an MRI exposes workers to a strong MF. The World Health Organization has now given the monitoring of occupational EMF exposure a high priority. One design for a low-cost, compact MF exposure monitor (« MR exposimeter ») uses a set of three orthogonally assembled Hall sensors. However, at such a strong EMF exposure intensity, the non-linearity and non-orthogonality (misalignment between the three Hall sensors) have an impact on the accuracy of EMF measurement. Therefore, a sensor characterization was performed in order to link Hall-effect output voltage to MF intensity. The sensor was then calibrated using an orthogonalization matrix and an offset vector. For each sensor configuration, the matrix and vector parameters were optimized with a calibration set generated by the movement of a three-axis sensor inside homogeneous MF areas. Once calibrated, the sensor was tested at different MF intensities and returned accuracy improvements. This calibration procedure was tested on synthetic data and performed on experimental data. The calibration parameters can be easily reused by the user, and their stability could be used as a quality control sensor. Finally, real-time monitoring test for static MF exposure was completed and validated on an MRI worker during a typical working day. Bioelectromagnetics. 39:108-119, 2018. © 2018 Wiley Periodicals, Inc.

Sorti le 3 aout 2016 et applicable au premier janvier 2017, le decret 2016–1074 etablit des regles et recommandations a suivre pour proteger les travailleurs face aux risques dus aux champs electromagnetiques. Transposant la directive europeenne 2013/35/UE, il concerne tout employeur et impose au minimum l'etablissement d'une evaluation des risques. Les unites, clinique ou de recherche, utilisant des dispositifs d'imagerie par resonance magnetique (IRM) sont particulierement concernees. En effet, cette modalite d'imagerie emet des champs electromagnetiques susceptibles d'engendrer des niveaux d'exposition des travailleurs depassant les valeurs declenchant l'action et les valeurs limites d'exposition definies par la directive et le decret. Une section de ce dernier lui est d'ailleurs entierement consacree. Le present document propose, en premier lieu, une analyse detaillee du decret 2016–1074 et des obligations de l'employeur. Il propose, dans un second temps, une analyse concernant la mise en application de cette directive et du decret associe pour le dispositif particulier qu'est l'IRM.

International audience; Worker exposure to electromagnetic fields (EMF) is a growing concern of international commissions. A European directive from 2013 (2013/35/EU) recommend to estimate or measure EMF exposure of all exposed workers. Magnetic resonance imaging (MRI) workers are specially concerned by this point because they work all day long in the vicinity of a very strong magnet (generally 1.5 or 3 T), which cannot be turned off. Setting up a magnetic field monitoring device on these workers would therefore be a good way to ensure their security. European directive threshold adequacy could then be verified. But this verification does not ensure a complete analysis of the worker exposure. Such an analysis based on quality control charts and exposure time's metrics has been described in this paper. The proposed magnetic field exposure analysis has additionally been tested on a long-term exposure follow-up of 18 MRI workers during 2 months.

International audience; In this work we present an algorithm for synthesising pseudo rtMRI data of the vocal tract. rtMRI data on the midsagittal plane were used to synthesise target consonant-vowel (CV) using only a silence frame of the target speaker. For this purpose, several single speaker models were created. The input of the algorithm is a silence frame of both train and target speaker and the rtMRI data of the target CV. An image transformation is computed from each CV frame to the next one, creating a set of transformations that describe the dynamics of the CV production. Another image transformation is computed from the silence frame of train speaker to the silence frame of the target speaker and is used to adapt the set of transformations computed previously to the target speaker. The adapted set of transformations is applied to the silence of the target speaker tosynthesise his/her CV pseudo rtMRI data. Synthesised images from multiple single speaker models are frame aligned and then averaged to create the final version of synthesised images. Synthesised images are compared with the original ones using image cross-correlation. Results show good agreement between the synthesised and the original images.

International audience; We present an algorithm for augmenting the shape of the vocal tract using 3D static and 2D dynamic speech MRI data. While static 3D images have better resolution and provide spatial information, 2D dynamic images capture the transitions. The aim of this work is to combine strong points of these two types of data to obtain better image quality of 2D dynamic images and extend the 2D dynamic images to the 3D domain. To produce a 3D dynamic consonant-vowel (CV) sequence, our algorithm takes as input the 2D CV transition and the static 3D targets for C and V. To obtain the enhanced sequence of images , the first step is to find a transformation between the 2D images and the mid-sagittal slice of the acoustically corresponding 3D image stack, and then find a transformation between neighbouring sagittal slices in the 3D static image stack. Combination of these transformations allows producing the final set of images. In the present study we first examined the transformation from the 3D mid-sagittal frame to the 2D video in order to improve image quality and then we examined the extension of the 2D video to the 3rd dimension with the aim to enrich spatial information.

International audience; Objective: Despite being routinely acquired during MRI examinations for triggering or monitoring purposes, electrocardiogram (ECG) signal recording and analysis remain challenging due to the inherent magnetic environment of an MRI scanner. The ECG signals are particularly distorted by the induction of electrical fields in the body by the MRI gradients. In this study, we propose a new hardware and software solution for the acquisition of ECG signal during MRI up to 3 T.Approach: Instead of restricting the sensor bandwidth to limit these gradient artifacts, the new sensor architecture has a higher bandwidth, higher sampling frequency and larger input dynamics, in order to acquire the ECG signals and the gradient artifacts more precisely. Signal processing based on a novel detection algorithm and blanking are then applied for improved artifact suppression.Main results: The proposed sensor allows the gradient artifacts to be acquired more precisely, and these artifacts are recorded with peak-to-peak amplitudes two orders of magnitude larger than for QRS complexes. The proposed method outperforms a state-of-the-art approach both in terms of signal quality (+9% 'SNR') and accuracy of QRS detection (+11%).Significance: The proposed hardware and software solutions open the way for the acquisition of high-quality of ECG gating in MRI, and improved diagnostic quality of ECG signals in MRI.

International audience; In this work we describe the creation of ArtSpeechMRIfr: a real-time as well as static magnetic resonance imaging (rtMRI, 3D MRI) database of the vocal tract. The database contains also processed data: denoised audio, its phonetically aligned annotation, articulatory contours, and vocal tract volume information , which provides a rich resource for speech research. The database is built on data from two male speakers of French. It covers a number of phonetic contexts in the controlled part, as well as spontaneous speech, 3D MRI scans of sustained vocalic articulations, and of the dental casts of the subjects. The corpus for rtMRI consists of 79 synthetic sentences constructed from a phonetized dictionary that makes possible to shorten the duration of acquisitions while keeping a very good coverage of the phonetic contexts which exist in French. The 3D MRI includes acquisitions for 12 French vowels and 10 consonants, each of which was pronounced in several vocalic contexts. Ar-ticulatory contours (tongue, jaw, epiglottis, larynx, velum, lips) as well as 3D volumes were manually drawn for a part of the images.

International audience; Purpose: To quantitatively evaluate a superresolution technique for 3D, one-millimeter isotropic diffusion-weighted imaging (DWI) of the whole breasts.Methods: Isotropic 3D DWI datasets are obtained using a combination of (i) a readout-segmented diffusion-weighted-echo-planar imaging (DW-EPI) sequence (rs-EPI), providing high in-plane resolution, and (ii) a superresolution (SR) strategy, which consists of acquiring 3 datasets with thick slices (3 mm) and 1-mm shifts in the slice direction, and combining them into a 1 × 1 × 1-mm3 dataset using a dedicated reconstruction. Two SR reconstruction schemes were investigated, based on different regularization schemes: conventional Tikhonov or Beltrami (an edge-preserving constraint). The proposed SR strategy was compared to native 1 × 1 × 1-mm3 acquisitions (i.e. with 1-mm slice thickness) in 8 healthy subjects, in terms of signal-to-noise ratio (SNR) efficiency, using a theoretical framework, Monte Carlo simulations and region-of-interest (ROI) measurements, and image sharpness metrics. Apparent diffusion coefficient (ADC) values in normal breast tissue were also compared.Results: The SR images resulted in an SNR gain above 3 compared to native 1 × 1 × 1-mm3 using the same acquisition duration (acquisition gain 3 and reconstruction gain >1). Beltrami-SR provided the best results in terms of SNR and image sharpness. The ADC values in normal breast measured from Beltrami-SR were preserved compared to low-resolution images (1.91 versus 1.97 ×10-3 mm2 /s, P = .1).Conclusion: A combination of rs-EPI and SR allows 3D, 1-mm isotropic breast DWI data to be obtained with better SNR than a native 1-mm isotropic acquisition. The proposed DWI protocol might be of interest for breast cancer monitoring/screening without injection.

International audience; This paper demonstrates that one can model the power deposited at the electrode of an implanted insulated cable submitted to the radiofrequency field of a 1.5 T magnetic resonance imaging modality (64 MHz) with a transmission line model. This offers an alternative that is more related to physics to the usually used transfer function model. The equivalence between the models is shown through a finite difference model and a new analytical formula for the transfer function as a function of transmission line parameters. First, the possibility of modeling an insulated cable with a transmission line model was analyzed through full-wave numerical simulations. The assumption of a transmission line model underlying the transfer function model was shown to be right for a simple cable embedded in tissue imitating gel, and the transmission line parameters extracted from this analysis were consistent with analytical formulas derived from the laws of physics. The transmission line model predictions were first compared to experimental and simulated data of the cable transfer function and then to experimental and simulated data of the resonant behavior as a function of length of cables with different termination conditions. The measured and simulated transfer functions fit perfectly a transmission line model with an analytical expression of the propagating constant. The transmission line model extracted from the transfer function allows us to predict the resonant behavior of two cables with different termination conditions.

Exploiting redundancies between multiple images of an MRI examination can be formalized as the joint reconstruction of these images. The anatomy is preserved indeed so that specific constraints can be implemented (e.g. most of the features or spatial gradients should be in the same place in all these images) and only the contrast changes from one image to another need to be encoded. The application of this concept is particularly challenging in cardiovascular and body imaging due to the complex organ deformations, especially with the patient breathing. In this study a joint optimization framework is proposed for reconstructing multiple MR images together with a nonrigid motion model. The motion model takes into account both intra-image and inter-image motion and therefore can correct for most ghosting/blurring artifacts and misregistration between images. The framework was validated with free-breathing myocardial ${\rm T}_{2}$ mapping experiments from nine heart transplant patients at 1.5 T. Results showed improved image quality and excellent image alignment with the multi-image reconstruction compared to the independent reconstruction of each image. Segment-wise myocardial ${\rm T}_{2}$ values were in good agreement with the reference values obtained from multiple breath-holds (62.5 $\pm$ 11.1 ms against 62.2 $\pm$ 11.2 ms which was not significant with ${\rm p}=0.49$ ).

International audience; This work concerns the construction of articulatory models for synthesis of speech, and more specifically the velum and epiglottis. The direct application of principal component analysis to the contours of these articulators extracted from MRI images results in unrealistic factors due to delineation errors. The approach described in this paper relies on the application of PCA to the centerline of the articulator and a simple reconstruction algorithm to obtain the global articulator contour. The complete articulatory model was constructed from static Magnetic Resonance (MR) images because their quality is much better than that of dynamic MR images. We thus assessed the extent to which the model constructed from static images is capable of approaching the vocal tract shape in MR images recorded at 55 Hz for continuous speech. The analysis of reconstruction errors shows that it is necessary to add dynamic images to the database of static images, in particular to approach the tongue shape for the /l/ sound.

International audience; Purpose: Segmentation of cardiac cine MRI data is routinely used for the volumetric analysis of cardiac function. Conventionally, 2D contours are drawn on short-axis (SAX) image stacks with relatively thick slices (typically 8 mm). Here, an acquisition/reconstruction strategy is used for obtaining isotropic 3D cine datasets; reformatted slices are then used to optimize the manual segmentation workflow.Methods: Isotropic 3D cine datasets were obtained from multiple 2D cine stacks (acquired during free-breathing in SAX and long-axis (LAX) orientations) using nonrigid motion correction (cine-GRICS method) and super-resolution. Several manual segmentation strategies were then compared, including conventional SAX segmentation, LAX segmentation in three views only, and combinations of SAX and LAX slices. An implicit B-spline surface reconstruction algorithm is proposed to reconstruct the left ventricular cavity surface from the sparse set of 2D contours.Results: All tested sparse segmentation strategies were in good agreement, with Dice scores above 0.9 despite using fewer slices (3-6 sparse slices instead of 8-10 contiguous SAX slices). When compared to independent phase-contrast flow measurements, stroke volumes computed from four or six sparse slices had slightly higher precision than conventional SAX segmentation (error standard deviation of 5.4 mL against 6.1 mL) at the cost of slightly lower accuracy (bias of -1.2 mL against 0.2 mL). Functional parameters also showed a trend to improved precision, including end-diastolic volumes, end-systolic volumes, and ejection fractions).Conclusion: The postprocessing workflow of 3D isotropic cardiac imaging strategies can be optimized using sparse segmentation and 3D surface reconstruction. Magn Reson Med 79:2665-2675, 2018. © 2017 International Society for Magnetic Resonance in Medicine.

Object: To demonstrate the possibility to model the power deposited at the electrode of an insulated cable in gel at 64MHz with a transmission line model with a strictly passive electrode model. This offers a more related to physics alternative to the transfer function model. The equivalence between both models is shown.Materials and Methods: In a first step, the possibility of modeling an insulated cable with a transmission line model was analyzed through full-wave numerical simulations. An electrode model that has a physical meaning was proposed. The transmission line model predictions were confronted to experimental and simulated data of the transfer function of a cable and of the resonant behavior as a function of length of cables with different termination conditions.Results: The assumption of a transmission line model which underlies the transfer function model is right for a simple cable embedded in tissue imitating gel. A transmission line model extracted from a transfer function allows to predict the resonant behavior of two cables with different termination conditions.Conclusion: A transmission line model of an insulated cable embedded in tissue imitating gel offers a more directly related to physics alternative to the transfer function model.

International audience; The paper presents a method to acquire articulatory data from a sequence of MRI images at a high framerate. The acquisition rate is enhanced by partially collecting data in the kt-space. The combination of compressed sensing technique, along with homodyne reconstruction, enables the missing data to be recovered. The good reconstruction is guaranteed by an appropriate design of the sampling pattern. It is based on a pseudo-random Cartesian scheme, where each line is partially acquired for use of the homodyne reconstruction, and where the lines are pseudo-randomly sampled: central lines are constantly acquired and the sampling density decreases as the lines are far from the center. Application on real speech data show that the framework enables dynamic sequences of vocal tract images to be recovered at a framerate higher than 30 frames per second and with a spatial resolution of 1 mm. A method to extract articulatory data from contour identification is presented. It is intended, in fine, to be used for the creation of a large database of articulatory data.

Purpose: To assess cardiac function by means of a novel free-breathing cardiac magnetic resonance imaging (MRI) strategy. Materials and Methods: A stack of ungated 2D steady-state free precession (SSFP) slices was acquired during free breathing and reconstructed as cardiac cine imaging based on the generalized reconstruction by inversion of coupled systems (GRICS). A motion-compensated sliding window approach allows reconstructing cine movies with most motion artifacts cancelled. The proposed reconstruction uses prior knowledge from respiratory belts and electrocardiogram recordings and features a piecewise linear model that relates the electrocardiogram signal to cardiac displacements. The free-breathing protocol was validated in six subjects against a standard breath-held protocol. Results: Image sharpness, as assessed by the image gradient entropy, was comparable to that of breath-held images and significantly better than in uncorrected images. Volumetric parameters of cardiac function in the left ventricle (LV) and right ventricle (RV) were similar, including end-systolic volumes, end-diastolic volumes and mass, stroke volumes, and ejection fractions (with differences of 3% ± 2.4 in the LV and 2.9% ± 4.4 in the RV). The duration of the free-breathing protocol was nearly the same as the breath-held protocol. Conclusion: Free-breathing cine-GRICS enables accurate assessment of volumetric parameters of cardiac function with efficient correction of motion. J. Magn. Reson. Imaging 2012;340-351. © 2011 Wiley Periodicals, Inc.

Black Blood Fast Spin Echo imaging of the heart is usually performed during mid-diastolic rest. This is a direct consequence of the long inversion time required to suppress the blood signal, which is constrained by the T(1) of the blood, and of the heart rate. To overcome these constraints, and to acquire black blood images in the end-systolic rest period, a new approach is introduced aiming at adaptively predicting the best time to prepare and acquire MR signals. It is based on a RR interval prediction algorithm and on a cardiac cycle model. The proposed method was applied to 14 healthy volunteers and is compared to a simple alternative method using a fixed delay and to the standard black blood imaging method for imaging in the mid-diastolic rest period. Results show that the proposed method offers an increased robustness in terms of trigger delay error and image quality compared to the tested simple alternative. Also, it has been shown by qualitative analysis done by an experienced observer that the right ventricle, especially the thin right ventricle free wall, is better depicted with our method than with the standard mid-diastolic rest acquisition.

This paper describes a rigorous framework for reconstructing MR images of the heart, acquired continuously over the cardiac and respiratory cycle. The framework generalizes existing techniques, commonly referred to as retrospective gating, and is based on the properties of reproducing kernel Hilbert spaces. The reconstruction problem is formulated as a moment problem in a multidimensional reproducing kernel Hilbert spaces (a two-dimensional space for cardiac and respiratory resolved imaging). Several reproducing kernel Hilbert spaces were tested and compared, including those corresponding to commonly used interpolation techniques (sinc-based and splines kernels) and a more specific kernel allowed by the framework (based on a first-order Sobolev RKHS). The Sobolev reproducing kernel Hilbert spaces was shown to allow improved reconstructions in both simulated and real data from healthy volunteers, acquired in free breathing.

A reconstruction strategy is proposed for physiological motion correction, which overcomes many limitations of existing techniques. The method is based on a general framework allowing correction for arbitrary motion–nonrigid or affine, making it suitable for cardiac or abdominal imaging, in the context of multiple coil, arbitrarily sampled acquisition. A model is required to predict motion in the field of view at each sample time point, based on prior knowledge provided by external sensors. A theoretical study is carried out to analyze the influence of motion prediction errors. Small errors are shown to propagate linearly in that reconstruction algorithm, and thus induce a reconstruction residue that is bounded (stability). Furthermore, optimization of the motion model is proposed in order to minimize this residue. This leads to reformulating reconstruction as two inverse problems which are coupled: motion-compensated reconstruction (known motion) and model optimization (known image). A fixed-point multiresolution scheme is described for inverting these two coupled systems. This framework is shown to allow fully autocalibrated reconstructions, as coil sensitivities and motion model coefficients are determined directly from the corrupted raw data. The theory is validated with real cardiac and abdominal data from healthy volunteers, acquired in free-breathing. Magn Reson Med 60:146–157, 2008. © 2008 Wiley-Liss, Inc.

Background During pediatric cardiac Cine-MRI, data acquired during cycles of different lengths must be combined. Feinstein’s model is used to project multiple cardiac cycles of variable lengths into a mean cycle. Methods 1/The temporal errors during Feinstein’s projection were computed in 306 cardiac cycles fully characterized by tissue Doppler imaging with 6-phase analysis (from a population of 7 children and young adults). 2/The effects of these temporal errors on tissue velocities were assessed by simulating a typical tissue phase mapping acquisition and reconstruction. 3/Myocardial velocities curves, extracted from high-resolution phase-contrast cine images, were compared for the 6 volunteers with lowest and highest heart rate variability, within a population of 36young adults. Results 1/The mean of temporal misalignments was 30ms over the cardiac cycle but reached 60ms during early diastole. 2/During phase contrast MRI simulation, early diastole velocity peaks were diminished by 5cm/s leading to virtual disappearance of isovolumic relaxation peaks. 3/The smoothing and erasing of isovolumic relaxation peaks was confirmed on tissue phase mapping velocity curves, between subjects with low and high heart rate variability (p=0.05). Conclusions Feinstein cardiac model creates temporal misalignments that impair high temporal resolution (notably for phase contrast cine imaging) in a population of high heart rate variations such as in paediatry (figure next page). Download : Download high-res image (187KB) Download : Download full-size image Abstract 0217 – Figure: Velocity curves with/without heart rate variation

Exploiting redundancies between multiple images of an MRI examination can be formalized as the joint reconstruction of these images. The anatomy is preserved indeed so that specific constraints can be implemented (e.g. most of the features or spatial gradients should be in the same place in all these images) and only the contrast changes from one image to another need to be encoded. The application of this concept is particularly challenging in cardiovascular and body imaging due to the complex organ deformations, especially with the patient breathing. In this study a joint optimization framework is proposed for reconstructing multiple MR images together with a nonrigid motion model. The motion model takes into account both intra-image and inter-image motion and therefore can correct for most ghosting/blurring artifacts and misregistration between images. The framework was validated with free-breathing myocardial T2 mapping experiments from nine heart transplant patients at 1.5 T. Results showed improved image quality and excellent image alignment with the multi-image reconstruction compared to the independent reconstruction of each image. Segment-wise myocardial T2 values were in good agreement with the reference values obtained from multiple breath-holds (62.5 ± 11.1 ms against 62.2 ± 11.2 ms which was not significant with p=0.49).

Introduction MRI is nowadays an indispensable technique with its excellent resolution achieved without ionizing radiations. Active implantable medical devices exist (such as pacemakers) which are MRI-compatible and do not show any hazardous local heating when interacting with the MRI machine radiofrequency, but multiple devices, as when two leads are side by side, were studied [1] , [2] but are not know enough to allow the implanted patients in these cases to undergo an MRI scan. The work presented here focuses on the coupling effects and the resulting heating between two leads, by simulation and experiments. Material and methods An ASTM [3] phantom inside a radiofrequency “birdcage” antenna was modeled with the CST Microwave Studio software (CST GmbH, Darmstadt, Germany). Two simples metallic cables, isolated and 5 mm bare at one tip were inserted in the phantom, spaced with 2.5 mm. One of them was also isolated at the upper extremity (referred to as “capped”) while the other one had its two tips bare (referred to as “uncapped”). An electromagnetic simulation was made and its results were used to launch a thermal simulation. With the latter, the temperature variations versus time were collected at both cables’ tips. The same process was repeated for each cable alone. Corresponding experiment were made with a Signa HDx 1.5 T MRI machine (GE Healthcare Technologies, Milwaukee, WI), an ASTM phantom filled with gel and cables made to match the simulations. An FSE sequence was used, generating high radiofrequency heating measured by optical temperature probes. Results Simulations and experiments show a coupling between the two cables: it is indeed observed that the “capped” cable heating is halved in presence of the “uncapped” cable. It is also observed that alone, the “capped” cable heats more that the “uncapped” one, but once together the higher heating is at the “uncapped” cable’s tip (see the appendix for the curves). Conclusion This study highlights the existence of lead coupling, such as it could occur in a patient’s body. This coupling could be explained by a theoretical model of two antennas subjected to radiofrequency, including mutual inductance and capacitance between the cables. Moreover, experiments on implant leads such as pacemaker leads could allow concluding on the occurrence of the phenomenon in more concrete cases. In the end, understanding this kind of interactions will allow a evolution in the design of medical devices, for each patient to be allowed to undergo an MRI scan safely, whatever his or her implants. Acknowledgement The authors thank the Region Lorraine and FEDER for financial support. Download : Download high-res image (188KB) Download : Download full-size image

International audience; Cardiac cine MRI with 3D isotropic resolution is challenging as it requires efficient data acquisition and motion management. It is proposed to use a 2D balanced SSFP (steady-state free precession) sequence rather than its 3D version as it provides better contrast between blood and tissue. In order to obtain 3D isotropic images, 2D multi-slice datasets are acquired in different orientations (short axis, horizontal long axis and vertical long axis) while the patient is breathing freely. Image reconstruction is performed in two steps: (i) a motion-compensated reconstruction of each image stack corrects for nonrigid cardiac and respiratory motion; (ii) a super-resolution (SR) algorithm combines the three motion-corrected volumes (with low resolution in the slice direction) into a single volume with isotropic resolution. The SR reconstruction was implemented with two regularization schemes including a conventional one (Tikhonov) and a feature-preserving one (Beltrami). The method was validated in 8 volunteers and 10 patients with breathing difficulties. Image sharpness, as assessed by intensity profiles and by objective metrics based on the structure tensor, was improved with both SR techniques. The Beltrami constraint provided efficient denoising without altering the effective resolution.

International audience; BackgroundDuring paediatric cardiac Cine-MRI, data acquired during cycles of different lengths must be combined. Most of the time, Feinstein’s model is used to project multiple cardiac cycles of variable lengths into a mean cycle.ObjectiveTo assess the effect of Feinstein projection on temporal resolution of Cine-MRI.Methods1/The temporal errors during Feinstein’s projection were computed in 306 cardiac cycles fully characterized by tissue Doppler imaging with 6-phase analysis (from a population of 7 children and young adults). 2/The effects of these temporal errors on tissue velocities were assessed by simulating typical tissue phase mapping acquisitions and reconstructions. 3/Myocardial velocities curves, extracted from high-resolution phase-contrast cine images, were compared for the 6 volunteers with lowest and highest heart rate variability, within a population of 36 young adults.Results1/The mean of temporal misalignments was 30 ms over the cardiac cycle but reached 60 ms during early diastole. 2/During phase contrast MRI simulation, early diastole velocity peaks were diminished by 6.1 cm/s leading to virtual disappearance of isovolumic relaxation peaks. 3/The smoothing and erasing of isovolumic relaxation peaks was confirmed on tissue phase mapping velocity curves, between subjects with low and high heart rate variability (p = 0.05).ConclusionsFeinstein cardiac model creates temporal misalignments that impair high temporal resolution phase contrast cine imaging when beat-to-beat heart rate is changing.

International audience; PURPOSE:To evaluate if measurement of split renal function ( SRF split renal function ) with dynamic contrast material-enhanced ( DCE dynamic contrast enhanced ) magnetic resonance (MR) urography is equivalent to that with renal scintigraphy ( RS renal scintigraphy ) in patients suspected of having chronic urinary obstruction.MATERIALS AND METHODS:The study protocol was approved by the institutional ethics committee of the coordinating center on behalf of all participating centers. Informed consent was obtained from all adult patients or both parents of children. This prospective, comparative study included 369 pediatric and adult patients from 14 university hospitals who were suspected of having chronic or intermittent urinary obstruction, and data from 295 patients with complete data were used for analysis. SRF split renal function was measured by using the area under the curve and the Patlak-Rutland methods, including successive review by a senior and an expert reviewer and measurement of intra- and interobserver agreement for each technique. An equivalence test for mean SRF split renal function was conducted with an α of 5%.RESULTS:Reproducibility was substantial to almost perfect for both methods. Equivalence of DCE dynamic contrast enhanced MR urography and RS renal scintigraphy for measurement of SRF split renal function was shown in patients with moderately dilated kidneys (P < .001 with the Patlak-Rutland method). However, in severely dilated kidneys, the mean SRF split renal function measurement was underestimated by 4% when DCE dynamic contrast enhanced MR urography was used compared with that when RS renal scintigraphy was used. Age and type of MR imaging device had no significant effect.CONCLUSION:For moderately dilated kidneys, equivalence of DCE dynamic contrast enhanced MR urography to RS renal scintigraphy was shown, with a standard deviation of approximately 12% between the techniques, making substitution of DCE dynamic contrast enhanced MR urography for RS renal scintigraphy acceptable. For severely dilated kidneys, a mean underestimation of SRF split renal function of 4% should be expected with DCE dynamic contrast enhanced MR urography, making substitution questionable.

International audience; OBJECT:Phase contrast imaging is widely used to measure blood velocity. However tissue Doppler imaging (TDI) echocardiography is the reference for myocardial velocity assessment. This study aims at validating the ability of phase contrast (PC) sequences to correctly assess myocardial velocities and to compare these velocities to TDI. The phase contrast sequence was performed with breath-hold parameters and with parameters tuned to increase temporal resolution in free breathing.MATERIALS AND METHODS:Left and Right auriculo-ventricular annuluses longitudinal velocities were recorded on six healthy volunteers with different temporal resolutions (TDI: 5 ms, breath-hold PC: 94 ms and free-breathing PC: 19 ms). Free-breathing PC was obtained by averaging of three excitations. Amplitudes of four standard echocardiographic and clinically relevant myocardial longitudinal velocity waves were compared: Early filling and auricular, systolic and isovolumic contractions.RESULTS:Isovolumic contraction waves were only visible with free-breathing PC and TDI. The differences with the reference TDI wave velocities were lower (p = 0.02) for free-breathing PC (19.2 ± 2.6%) than for breath-hold PC (28.1 ± 2.9%). These differences for free-breathing PC were close to (p = 0.21) the coefficient of variation of the measurements provided by TDI (14.8 ± 1.2%).CONCLUSION:Myocardial longitudinal peak velocities can be assessed with a PC sequence tuned to optimize temporal resolution.

International audience; Aim: The aim of this work was the development of successful cell therapy techniques for cartilage engineering. This will depend on the ability to monitor non-invasively transplanted cells, especially mesenchymal stem cells (MSCs) that are promising candidates to regenerate damaged tissues. Methods: MSCs were labeled with superparamagnetic iron oxide particles (SPIO). We examined the effects of long-term labeling, possible toxicological consequences and the possible influence of progressive concentrations of SPIO on chondrogenic differentiation capacity. Results: No influence of various SPIO concentrations was noted on human bone marow MSC viability or proliferation. We demonstrated long-term (4 weeks) in vitro retention of SPIO by human bone marrow MSCs seeded in collagenic sponges under TGF-beta 1 chondrogenic conditions, detectable by Magnetic Resonance Imaging (MRI) and histology. Chondrogenic differentiation was demonstrated by molecular and histological analysis of labeled and unlabeled cells. Chondrogenic gene expression (COL2A2, ACAN, SOX9, COL10, COMP) was significantly altered in a dose-dependent manner in labeled cells, as were GAG and type II collagen staining. As expected, SPIO induced a dramatic decrease of MRI T2 values of sponges at 7T and 3T, even at low concentrations. Conclusions: This study clearly demonstrates (1) long-term in vitro MSC traceability using SPIO and MRI and (2) a deleterious dose-dependence of SPIO on TGF-beta 1 driven chondrogenesis in collagen sponges. Low concentrations (12.5-25 mu g Fe/mL) seem the best compromise to optimize both chondrogenesis and MRI labeling.

To validate a new index, the surface-length index (SLI) based on area change in a short-axis view and length reduction in the horizontal long-axis view, which is used to quickly (1 min) detect right ventricles with an abnormal ejection fraction (EF) during a cardiac MRI examination. SLI can be used to avoid a complete delineation of the endocardial contours of normal right ventricles.Sixty patients (group A) were retrospectively included to calibrate the SLI formula by optimisation of the area under the ROC curves and SLI thresholds were chosen to obtain 100 % sensitivity. Another 340 patients (group B) were prospectively recruited to test SLI's capacity to detect right ventricles (RVs) with an abnormal EF (0.5).The appropriate threshold to obtain 100 % sensitivity in group A was 0.58. In group B, with the 0.58 threshold, SLI yielded a sensitivity of 100 % and specificity of 51 %. SLI would have saved 35 % of the RV studies in our population, without inducing any diagnostic error. SLI and EF correlation was good (r (2) = 0.64).SLI combines two simple RV measures, and brings significant improvement in post-processing efficiency by preselecting RVs that require a complete study.• Assessment of right ventricle ejection fraction (RVEF) with cine-MRI is time consuming. • Therefore, RVEF is not always assessed during cardiac MRI. • Surface-length index (SLI) allows rapid detection of abnormal RVEF during cardiac MRI. • SLI saves one third of the operator time. • Every cardiac MRI could include RVEF assessment by means of SLI.

International audience; Non-invasive quantitative assessment of articular cartilage integrity is essential for early detection and evaluation of osteoarthritis (OA) and for the follow-up of stem-cell-driven cartilage engineering. In this study, we investigated the feasibility of exploiting diffusion tensor imaging (DTI) on porcine knee joints with a clinical magnetic resonance (MR) scanner to extract micro-structural information in order to complement biochemical information quantified by T2 maps. We propose an MR protocol for quantifying T2 and cartilage microstructure with diffusion MR on a clinical scanner. Preliminary results were obtained on four pig knee joints using a 3 T GE clinical MRI scanner and an 8-channel knee coil array. The measured cartilage volume, T2 values, apparent diffusion coefficient and fractional anisotropy (FA) of femoral and tibial cartilage were respectively 9.8/2.3 mm2, 67.0/56.1 ms, 1.3/1.3×10-3 mm2/s and 0.4/0.3. This new protocol has the potential to be combined in vivo with quantitative assessment of both cartilage degradation and restoration in osteoarthritis.

Magnetic resonance images are affected by motion artefacts due to breathing and cardiac beating that occur during the acquisition. Methods for joint reconstruction of image and motion have been proposed recently. Such optimization problems are ill-conditioned, therefore regularization methods are required such as motion smoothness constraints using the Tikhonov method. However with Tikhonov methods the solution often relies on a good choice of the regularization parameter micron, especially in large parameter search spaces (e.g., in 3D reconstructions). In this paper, we propose an adaptive, implicit regularization method which results in subject-specific, spatially varying smoothness constraints on the motion model. It is based on the idea of solving for motion only in certain key points that form a mesh. A practical algorithm is proposed for generating this mesh automatically. The proposed method is shown to have a better convergence rate than the Tikhonov method, both in silico and in vivo. The accuracy of the reconstructed image and motion is also improved.

Object Quantitative T 2 measurement is applied in cardiac Magnetic Resonance Imaging (MRI) for the diagnosis and follow-up of myocardial pathologies. Standard Electrocardiogram (ECG)-gated fast spin echo pulse sequences can be used clinically for T 2 assessment, with multiple breath-holds. However, heart rate is subject to physiological variability, which causes repetition time variations and affects the recovery of longitudinal magnetization between TR periods. Materials and methods The bias caused by heart rate variability on quantitative T 2 measurements is evaluated for fast spin echo pulse sequence. Its retrospective correction based on an effective TR is proposed. Heart rate variations during breath-holds are provided by the ECG recordings from healthy volunteers. T 2 measurements were performed on a phantom with known T 2 values, by synchronizing the sequence with the recorded ECG. Cardiac T 2 measurements were performed twice on six volunteers. The impact of T 1 on T 2 is also studied. Results Maximum error in T 2 is 26% for phantoms and 18% for myocardial measurement. It is reduced by the proposed compensation method to 20% for phantoms and 10% for in vivo measurements. Only approximate knowledge of T 1 is needed for T 2 correction. Conclusion Heart rate variability may cause a bias in T2 measurement with ECG-gated FSE. It needs to be taken into account to avoid a misleading diagnosis from the measurements.

Summary This study demonstrates the feasibility of free-breathing phase-contrast CINE MRI without averaging. A new version of the CINE GRICS algorithm[1] was used to correct for motion. Background Phase-contrast MRI encodes speed and direction of moving spins by means of toggling a bipolar gradient. It is a valuable tool for assessing conditions affecting the vascular system by measuring the velocity of flowing blood[2]. Clinically, this sequence is performed in breath-hold or in free breathing but, in the latter case, using signal averaging. We propose to demonstrate the feasibility of free-breathing phase-contrast CINE MRI without averaging exploiting the acquisition redundancy by applying a new version of the CINE GRICS algorithm[1] to correct for motion. Methods Cardiac examination (approved by our local ethics committee) was performed on one normal volunteer during which three 2D phase-contrast CINE MRI sequences (common parameters: 256x128 acquisition matrix, 6 views per segment (vps), 32 reconstructed cardiac phases, 150 cm/sec VENC, slice direction velocity encoding, 5 mm slice thickness, 44 cm FOV, 62.5 kHz bandwidth, 3.05/8.08 ms TE/TR, 15° flip angle) were acquired on a 3T scanner (Signa HDxt, GE Healthcare, Milwaukee, WI) with a 8-element cardiac coil : (1) breath-held (2) averaged (3 NEX) in free breathing (3) in free breathing storing the raw data of 3 NEX to an external computer for offline processing. Signals from a respiratory belt were carried by a custom Maglife patient monitoring system (Schiller Medical, France) and recorded with a dedicated home-made hardware. Offline processing consisted of splitting the raw data from the 2 velocity encoding steps thus giving 2 sets of

Magnetic resonance images are affected by motion artefacts due to breathing and cardiac beating that occur during the acquisition. Methods for joint reconstruction of image and motion have been proposed recently. Such optimization problems are ill-conditioned, therefore regularization methods are required such as motion smoothness constraints using the Tikhonov method. However with Tikhonov methods the solution often relies on a good choice of the regularization parameter μ, especially in large parameter search spaces (e.g. in 3D reconstructions). In this paper, we propose an adaptive, implicit regularization method which results in subject-specific, spatially varying smoothness constraints on the motion model. It is based on the idea of solving for motion only in certain key points that form a mesh. A practical algorithm is proposed for generating this mesh automatically. The proposed method is shown to have a better convergence rate than the Tikhonov method, both in silico and in vivo. The accuracy of the reconstructed image and motion is also improved.

International audience; Reproducibility of the manual assessment of right ventricle volumes by short-axis cine-MRI remains low and is often attributed to the difficulty in separating the right atrium from the ventricle. This study was designed to evaluate the regional interobserver variability of the right ventricle volume assessment to identify segmentation zones with the highest interobserver variability. Short-axis views of 90 right ventricles (30 hypertrophic, 30 dilated, and 30 normal) were acquired with 2D steady-state free precession sequences at 1.5 T and were manually segmented by two observers. The two segmentations were compared and the variations were quantified with a variation score based on the Hausdorff distance between the two segmentations and the interobserver 95% limits of concordance of the global volumes. The right ventricles were semiautomatically split into four subregions: apex, mid-ventricle, tricuspid zone, and infundibulum. These four subregions represented 11%, 34%, 36%, and 19% of the volume but, respectively, yielded variation scores of 8%, 16%, 42%, and 34%. The infundibulum yielded the highest interobserver regional variability although its variation score remained comparable to the tricuspid zone due to its lower volume. These results emphasize the importance of standardizing the segmentation of the infundibulum and the tricuspid zone to improve reproducibility.

The analysis of abdominal and thoracic dynamic contrast-enhanced MRI is often impaired by artifacts and misregistration caused by physiological motion. Breath-hold is too short to cover long acquisitions. A novel multipurpose reconstruction technique, entitled dynamic contrast-enhanced generalized reconstruction by inversion of coupled systems, is presented. It performs respiratory motion compensation in terms of both motion artefact correction and registration. It comprises motion modeling and contrast-change modeling. The method feeds on physiological signals and x-f space properties of dynamic series to invert a coupled system of linear equations. The unknowns solved for represent the parameters for a linear nonrigid motion model and the parameters for a linear contrast-change model based on B-splines. Performance is demonstrated on myocardial perfusion imaging, on six simulated data sets and six clinical exams. The main purpose consists in removing motion-induced errors from time–intensity curves, thus improving curve analysis and postprocessing in general. This method alleviates postprocessing difficulties in dynamic contrast-enhanced MRI and opens new possibilities for dynamic contrast-enhanced MRI analysis. Magn Reson Med, 2011. © 2010 Wiley-Liss, Inc.

International audience; In clinical applications, cardiac gated sequences are commonly used for heart imaging. In practice, we simply wait a fixed time called Trigger Delay (TD) after the R-wave has been detected to acquire data in the chosen cardiac phase. The main problem of this method is that the TD is constant and does not take into account the physiological variability such as instantaneous heart rate changes during breath-hold or free breathing. Another issue is that we cannot use a TD shorter than the preparation time. Consequently when a cardiac gated sequence with a long preparation time is used, only diastolic images are achievable. Double Inversion Recovery Fast Spin Echo (DIR-FSE), resulting in black-blood images of the heart [1], is such a sequence since the inversion time (TI) needed to cancel the blood signal is around 500ms. Systolic view of the heart could also be of clinical interest, especially during end-systolic phase when the cardiac volume is minimum and constant (average TD = 300ms, average duration 60ms [2]). The primary aim of this work was to assess a robust method to acquire black blood FSE in end-systolic phase. For this purpose, the DIR preparation has to be launched before the R-wave in the previous cardiac cycle. A RR interval prediction is then needed and heart rate variability has to be accounted in order to position the acquisition window properly. A new general adaptive method is described here that overcomes the above listed limitations in prospectively cardiac gated sequences with long preparation time using a predictive model. This method has been applied on end-systolic black blood FSE images.

L'imagerie par résonance magnétique (IRM) du coeur reste difficile. Les solutions retenues dans la pratique clinique pour limiter les problèmes liés aux mouvements sont l'apnée et la synchronisation cardiaque. En imagerie cardiaque, la séquence Double Inversion Recovery Fast Spin Echo (DIR-FSE) est couramment utilisée [1]. Cependant, cette séquence, synchronisée avec l'activité cardiaque, nécessite un temps de préparation long (TI). Le temps entre l'onde R et l'acquisition (TD) est donc également long. Par conséquent, cette séquence n'est utilisée qu'en diastole (figure 1-a). De plus, les variations de rythme cardiaque instantanées (temps entre 2 ondes R consécutives) ne sont pas prises en compte. Afin de passer outre ces limitations, nous proposons deux techniques pour acquérir les séquences DIR-FSE en fin de systole (relaxation iso-volumétrique, volume cardiaque minimum et constant). Ces acquisitions en fin de systole devraient nous permettre de mieux voir le ventricule droit en raison de sa contraction. Les deux méthodes proposées seront comparées entre elles en terme de robustesse. Elles seront également comparées à la méthode conventionnelle en terme de visualisation du ventricule droit.

International audience; This article describes a general framework for multiple coil MRI reconstruction in the presence of elastic physiological motion. On the assumption that motion is known or can be predicted, it is shown that the reconstruction problem is equivalent to solving an integral equation--known in the literature as a Fredholm equation of the first kind--with a generalized kernel comprising Fourier and coil sensitivity encoding, modified by physiological motion information. Numerical solutions are found using an iterative linear system solver. The different steps in the numerical resolution are discussed, in particular it is shown how over-determination can be used to improve the conditioning of the generalized encoding operator. Practical implementation requires prior knowledge of displacement fields, so a model of patient motion is described which allows elastic displacements to be predicted from various input signals (e.g., respiratory belts, ECG, navigator echoes), after a free-breathing calibration scan. Practical implementation was demonstrated with a moving phantom setup and in two free-breathing healthy subjects, with images from the thoracic-abdominal region. Results show that the method effectively suppresses the motion blurring/ghosting artifacts, and that scan repetitions can be used as a source of over-determination to improve the reconstruction.