Compensation for respiratory motion-induced signal loss and phase corruption in free-breathing self-navigated cine DENSE using deep learning.

PURPOSE: To introduce a model that describes the effects of rigid translation due to respiratory motion in displacement encoding with stimulated echoes (DENSE) and to use the model to develop a deep convolutional neural network to aid in first-order respiratory motion compensation for self-navigated free-breathing cine DENSE of the heart. METHODS: The motion model includes conventional position shifts of magnetization and further describes the phase shift of the stimulated echo due to breathing. These image-domain effects correspond to linear and constant phase errors, respectively, in k-space. The model was validated using phantom experiments and Bloch-equation simulations and was used along with the simulation of respiratory motion to generate synthetic images with phase-shift artifacts to train a U-Net, DENSE-RESP-NET, to perform motion correction. DENSE-RESP-NET-corrected self-navigated free-breathing DENSE was evaluated in human subjects through comparisons with signal averaging, uncorrected self-navigated free-breathing DENSE, and breath-hold DENSE. RESULTS: Phantom experiments and Bloch-equation simulations showed that breathing-induced constant phase errors in segmented DENSE leads to signal loss in magnitude images and phase corruption in phase images of the stimulated echo, and that these artifacts can be corrected using the known respiratory motion and the model. For self-navigated free-breathing DENSE where the respiratory motion is not known, DENSE-RESP-NET corrected the signal loss and phase-corruption artifacts and provided reliable strain measurements for systolic and diastolic parameters. CONCLUSION: DENSE-RESP-NET is an effective method to correct for breathing-associated constant phase errors. DENSE-RESP-NET used in concert with self-navigation methods provides reliable free-breathing DENSE myocardial strain measurement.

High frame rate deformation analysis of knee cartilage by spiral dualMRI and relaxation mapping.

PURPOSE: Daily activities including walking impose high-frequency cyclic forces on cartilage and repetitive compressive deformation. Analyzing cartilage deformation during walking would provide spatial maps of displacement and strain and enable viscoelastic characterization, which may serve as imaging biomarkers for early cartilage degeneration when the damage is still reversible. However, the time-dependent biomechanics of cartilage is not well described, and how defects in the joint impact the viscoelastic response is unclear. METHODS: We used spiral acquisition with displacement-encoding MRI to quantify displacement and strain maps at a high frame rate (25 frames/s) in tibiofemoral joints. We also employed relaxometry methods (T1 , T1ρ , T2 , T2 *) on the cartilage. RESULTS: Normal and shear strains were concentrated on the bovine tibiofemoral contact area during loading, and the defected joint exhibited larger compressive strains. We also determined a positive correlation between the change of T1ρ in cartilage after cyclic loading and increased compressive strain on the defected joint. Viscoelastic behavior was quantified by the time-dependent displacement, where the damaged joint showed increased creep behavior compared to the intact joint. This technique was also successfully demonstrated on an in vivo human knee showing the gradual change of displacement during varus load. CONCLUSION: Our results indicate that spiral scanning with displacement encoding can quantitatively differentiate the damaged from intact joint using the strain and creep response. The viscoelastic response identified with this methodology could serve as biomarkers to detect defects in joints in vivo and facilitate the early diagnosis of joint diseases such as osteoarthritis.

Multi-frame biomechanical and relaxometry analysis during in vivo loading of the human knee by spiral dualMRI and compressed sensing.

PURPOSE: Knee cartilage experiences repetitive loading during physical activities, which is altered during the pathogenesis of diseases like osteoarthritis. Analyzing the biomechanics during motion provides a clear understanding of the dynamics of cartilage deformation and may establish essential imaging biomarkers of early-stage disease. However, in vivo biomechanical analysis of cartilage during rapid motion is not well established. METHODS: We used spiral displacement encoding with stimulated echoes (DENSE) MRI on in vivo human tibiofemoral cartilage during cyclic varus loading (0.5 Hz) and used compressed sensing on the k-space data. The applied compressive load was set for each participant at 0.5 times body weight on the medial condyle. Relaxometry methods were measured on the cartilage before (T1ρ , T2 ) and after (T1ρ ) varus load. RESULTS: Displacement and strain maps showed a gradual shift of displacement and strain in time. Compressive strain was observed in the medial condyle cartilage and shear strain was roughly half of the compressive strain. Male participants had more displacement in the loading direction compared to females, and T1ρ values did not change after cyclic varus load. Compressed sensing reduced the scanning time up to 25% to 40% when comparing the displacement maps and substantially lowered the noise levels. CONCLUSION: These results demonstrated the ease of which spiral DENSE MRI could be applied to clinical studies because of the shortened imaging time, while quantifying realistic cartilage deformations that occur through daily activities and that could serve as biomarkers of early osteoarthritis.

In Vitro Validation of Regional Circumferential Strain Assessment in a Phantom Aortic Model Using Cine Displacement Encoding With Stimulated Echoes MRI.

BACKGROUND: A novel application of cine Displacement ENcoding with Stimulated Echoes Magnetic Resonance Imaging (DENSE MRI) has recently been described to assess regional heterogeneities in circumferential strain around the aortic wall in vivo; however, validation is first required for successful clinical translation. PURPOSE: To validate the quantification of regional circumferential strain around the wall of an aortic phantom using DENSE MRI. STUDY TYPE: In vitro phantom study. POPULATION: Three polyvinyl alcohol aortic phantoms with eight axially oriented nitinol wires embedded evenly around the walls. FIELD STRENGTH/SEQUENCE: 3 T; gradient-echo aortic DENSE MRI with spiral cine readout, gradient-echo phase-contrast MRI (PCMR) with Cartesian cine readout. ASSESSMENT: Phantoms were connected to a pulsatile flow loop and peak DENSE-derived regional circumferential Green strains at 16 equally spaced sectors around the wall were assessed according to previously published algorithms. "True" regional circumferential strains were calculated by manually tracking displacements of the nitinol wires by two independent observers. Normalized circumferential strains (NCS) were calculated by dividing regional strains by the mean strain. Finally, DENSE-derived regional strain was corrected by multiplying regional DENSE NCS by the mean strain calculated from the diameter change on the PCMR. STATISTICAL TESTS: One-sample t-test, Paired-sample t-test, and analysis of variance with Bonferroni correction, coefficient of variation (CoV), Bland-Altman analysis; P < 0.05 was considered statistically significant. RESULTS: Aortic DENSE MRI significantly overestimated circumferential strain compared to the wire-tracking method (mean difference and SD 0.030 ± 0.014, CoV 0.31). However, NCS demonstrated good agreement between DENSE and wire-tracking data (mean difference 0.000 ± 0.172, CoV 0.15). After correcting the DENSE-derived regional strain, the mean difference in regional circumferential strain between DENSE and wire-tracking was significantly reduced to 0.006 ± 0.008, and the CoV was reduced to 0.18. DATA CONCLUSION: For aortic phantoms with mild spatial heterogeneity in circumferential strain, the previously published aortic DENSE MRI technique successfully assessed the regional NCS distribution but overestimated the mean strain. This overestimation is correctable by computing a more accurate mean circumferential strain using a separate cine scan. LEVEL OF EVIDENCE: 2 TECHNICAL EFFICACY: Stage 2.

Assessment of the regional distribution of normalized circumferential strain in the thoracic and abdominal aorta using DENSE cardiovascular magnetic resonance.

BACKGROUND: Displacement Encoding with Stimulated Echoes (DENSE) cardiovascular magnetic resonance (CMR) of the aortic wall offers the potential to improve patient-specific diagnostics and prognostics of diverse aortopathies by quantifying regionally heterogeneous aortic wall strain in vivo. However, before regional mapping of strain can be used to clinically assess aortic pathology, an evaluation of the natural variation of normal regional aortic kinematics is required. METHOD: Aortic spiral cine DENSE CMR was performed at 3 T in 30 healthy adult subjects (range 18 to 65 years) at one or more axial locations that are at high risk for aortic aneurysm or dissection: the infrarenal abdominal aorta (IAA, n = 11), mid-descending thoracic aorta (DTA, n = 17), and/or distal aortic arch (DAA, n = 11). After implementing custom noise-reduction techniques, regional circumferential Green strain of the aortic wall was calculated across 16 sectors around the aortic circumference at each location and normalized by the mean circumferential strain for comparison between individuals. RESULTS: The distribution of normalized circumferential strain (NCS) was heterogeneous for all locations evaluated. Despite large differences in mean strain between subjects, comparisons of NCS revealed consistent patterns of strain distribution for similar groupings of patients by axial location, age, and/or mean displacement angle. NCS at local systole was greatest in the lateral/posterolateral walls in the IAAs (1.47 ± 0.27), medial wall in anteriorly displacing DTAs (1.28 ± 0.20), lateral wall in posteriorly displacing DTAs (1.29 ± 0.29), superior curvature in DAAs < 50 years-old (1.93 ± 0.22), and medial wall in DAAs > 50 years (2.29 ± 0.58). The distribution of strain was strongly influenced by the location of the vertebra and other surrounding structures unique to each location. CONCLUSIONS: Regional in vivo circumferential strain in the adult aorta is unique to each axial location and heterogeneous around its circumference, but can be grouped into consistent patterns defined by basic patient-specific metrics following normalization. The heterogeneous strain distributions unique to each group may be due to local peri-aortic constraints (particularly at the aorto-vertebral interface), heterogeneous material properties, and/or heterogeneous flow patterns. These results must be carefully considered in future studies seeking to clinically interpret or computationally model patient-specific aortic kinematics.

Correction of phase offset errors and quantification of background noise, signal-to-noise ratio, and encoded-displacement uncertainty on DENSE MRI for kinematics of the descending thoracic and abdominal aorta.

Displacement encoding with stimulated echoes (DENSE) MRI is a phase contrast technique that allows the encoding of tissue displacement into the phase of the magnetic resonance signal. Recent developments in this technique allow the imaging of relatively thin structures such as the aortic wall. Quantifying background noise associated to DENSE MRI is required to assess the uncertainty of derived displacement measurements and for the design and implementation of adequate noise-reduction techniques. Although noise and error management of cardiac DENSE MRI has been previously studied, developments for aortic applications are scarce. Herein, we evaluate the noise and uncertainty of DENSE MRI scans at three different locations along the descending aorta: the distal aortic arch (DAA), the descending thoracic aorta (DTA), and infrarenal abdominal aorta (IAA). Additionally, we analyze three datasets from in vitro validation experiments with polyvinyl alcohol phantoms. We implement and evaluate the effectiveness of an offset-error correction algorithm and noise filtering techniques on DENSE MRI for aortic motion applications. Our results show that the phase signal of pixels composing the static background was normally distributed, centered on average at 0.003 ± 0.02 rad and - 0.02 ± 0.024 rad for each phase directions, suggesting that background noise is random, isotropic, and DENSE MRI has little offset errors. However, background signal noise significantly increased with elapsed time of the cardiac cycle; and was spatially heterogeneous consistently increased towards the anterior space. Background noise showed no significant differences between the 3 aortic locations and the in vitro experiments. However, SNR depended on the displacement of the region of interest, in consequence it was found significantly larger at DAA (16.7 ± 8.5, p = 0.003) and DTA (15.4 ± 7.6, p = 0.008) than at the IAA (8.0 ± 4.1), but not significantly different than the SNR of in vitro experiments (8.0 ± 3.7), and had an overall average of 13 ± 7. The applied methods significantly reduced the offset error and effect of noise on the estimation of encoded displacements. Finally, this analysis suggests that the implemented DENSE MRI protocol is adequate to assess the motion of healthy human aortas. However, the relative effect of noise increased considerably on the analysis of an ageing or diseased aortas with impaired mobility, calling for further analyses on pathologically stiffened aortas.

Quantification of the heterogeneous effect of static and dynamic perivascular structures on patient-specific local aortic wall mechanics using inverse finite element modeling and DENSE MRI.

Recent studies have highlighted the relevance of perivascular interactions on aortic wall mechanics. Most of the approaches assume static perivascular structures; however, the beating heart dynamically displaces the neighboring aorta. We develop a model to account for the effect of periaortic interactions due to static and dynamic structures by prescribing a moving elastic foundation boundary condition (EFBC) embedded into an inverse finite element algorithm using in vivo displacements from 2D displacement encoding with stimulated echoes (DENSE) MRI as target data. We applied this method at three different locations of interest, the distal aortic arch (DAA), descending thoracic aorta (DTA), and infrarenal abdominal aorta (IAA) for a total of 27 cases in healthy humans. The model reproduces the target diastole-to-systole deformation and bulk displacement of the aortic wall with median displacement errors below 0.5mm. The EFBC showed good agreement with the location of anatomical features and was consistent among individuals of similar characteristics. Results show that an energy source acting on the adventitia is required to reproduce the displacements measured at the vicinity of the heart, but not at the abdomen. The average adventitial load as a percentage of the luminal pulse-pressure was found to increase with age and to decrease along the descending aorta, from 61% at the DAA to 37% at the DTA, and 30% at the IAA. This approach offers a patient-specific method to estimate in vivo adventitial loads and aortic wall stiffness, which can bring a better understanding of normal and pathological in vivo aortic function.

Regional Brain Tissue Displacement and Strain is Elevated in Subjects with Chiari Malformation Type I Compared to Healthy Controls: A Study Using DENSE MRI.

While the degree of cerebellar tonsillar descent is considered the primary radiologic marker of Chiari malformation type I (CMI), biomechanical forces acting on the brain tissue in CMI subjects are less studied and poorly understood. In this study, regional brain tissue displacement and principal strains in 43 CMI subjects and 25 controls were quantified using a magnetic resonance imaging (MRI) methodology known as displacement encoding with stimulated echoes (DENSE). Measurements from MRI were obtained for seven different brain regions-the brainstem, cerebellum, cingulate gyrus, corpus callosum, frontal lobe, occipital lobe, and parietal lobe. Mean displacements in the cerebellum and brainstem were found to be 106 and 64% higher, respectively, for CMI subjects than controls (p < .001). Mean compression and extension strains in the cerebellum were 52 and 50% higher, respectively, in CMI subjects (p < .001). Brainstem mean extension strain was 41% higher in CMI subjects (p < .001), but no significant difference in compression strain was observed. The other brain structures revealed no significant differences between CMI and controls. These findings demonstrate that brain tissue displacement and strain in the cerebellum and brainstem might represent two new biomarkers to distinguish between CMI subjects and controls.

A comprehensive comparison between shortest-path HARP refinement, SinMod, and DENSEanalysis processing tools applied to CSPAMM and DENSE images.

We addressed comprehensively the performance of Shortest-Path HARP Refinement (SP-HR), SinMod, and DENSEanalysis using 2D slices of synthetic CSPAMM and DENSE images with realistic contrasts obtained from 3D phantoms. The three motion estimation techniques were interrogated under ideal and no-ideal conditions (with MR induced artifacts, noise, and through-plane motion), considering several resolutions and noise levels. Under noisy conditions, and for isotropic pixel sizes of 1.5 mm and 3.0 mm in CSPAMM and DENSE images respectively, the nRMSE obtained for the circumferential and radial strain components were 10.7 ±â€¯10.8% and 25.5 ±â€¯14.8% using SP-HR, 11.9 ±â€¯2.5% and 29.3 ±â€¯6.5% using SinMod, and 6.4 ±â€¯2.0% and 18.2 ±â€¯4.6% using DENSEanalysis. Overall, the results showed that SP-HR tends to fail for large tissue motions, whereas SinMod and DENSEanalysis gave accurate displacement and strain field estimations, being the last which performed the best.