Evaluation of Deep Learning Estimation of Whole Heart Anatomy from Automated Cardiovascular Magnetic Resonance Short- and Long-Axis Analyses in UK Biobank.

AIMS: Standard methods of heart chamber volume estimation in cardiovascular magnetic resonance (CMR) typically utilize simple geometric formulae based on a limited number of slices. We aimed to evaluate whether an automated deep learning neural network prediction of 3D anatomy of all four chambers would show stronger associations with cardiovascular risk factors and disease than standard volume estimation methods in the UK Biobank. METHODS: A deep learning network was adapted to predict 3D segmentations of left and right ventricles (LV, RV) and atria (LA, RA) at ∼1mm isotropic resolution from CMR short and long axis 2D segmentations obtained from a fully automated machine learning pipeline in 4723 individuals with cardiovascular disease (CVD) and 5733 without in the UK Biobank. Relationships between volumes at end-diastole (ED) and end-systole (ES) and risk/disease factors were quantified using univariate, multivariate and logistic regression analyses. Strength of association between deep learning volumes and standard volumes was compared using the area under the receiving operator characteristic curve (AUC). RESULTS: Univariate and multivariate associations between deep learning volumes and most risk and disease factors were stronger than for standard volumes (higher R2 and more significant P values), particularly for sex, age, and body mass index. AUC for all logistic regressions were higher for deep learning volumes than standard volumes (p<0.001 for all four chambers at ED and ES). CONCLUSIONS: Neural network reconstructions of whole heart volumes had significantly stronger associations with cardiovascular disease and risk factors than standard volume estimation methods in an automatic processing pipeline.

Inversion recovery and saturation recovery pulmonary vein MR angiography using an image based navigator fluoro trigger and variable-density 3D cartesian sampling with spiral-like order.

Contrast enhanced pulmonary vein magnetic resonance angiography (PV CE-MRA) has value in atrial ablation pre-procedural planning. We aimed to provide high fidelity, ECG gated PV CE-MRA accelerated by variable density Cartesian sampling (VD-CASPR) with image navigator (iNAV) respiratory motion correction acquired in under 4 min. We describe its use in part during the global iodinated contrast shortage. VD-CASPR/iNAV framework was applied to ECG-gated inversion and saturation recovery gradient recalled echo PV CE-MRA in 65 patients (66 exams) using .15 mmol/kg Gadobutrol. Image quality was assessed by three physicians, and anatomical segmentation quality by two technologists. Left atrial SNR and left atrial/myocardial CNR were measured. 12 patients had CTA within 6 months of MRA. Two readers assessed PV ostial measurements versus CTA for intermodality/interobserver agreement. Inter-rater/intermodality reliability, reproducibility of ostial measurements, SNR/CNR, image, and anatomical segmentation quality was compared. The mean acquisition time was 3.58 ± 0.60 min. Of 35 PV pre-ablation datasets (34 patients), mean anatomical segmentation quality score was 3.66 ± 0.54 and 3.63 ± 0.55 as rated by technologists 1 and 2, respectively (p = 0.7113). Good/excellent anatomical segmentation quality (grade 3/4) was seen in 97% of exams. Each rated one exam as moderate quality (grade 2). 95% received a majority image quality score of good/excellent by three physicians. Ostial PV measurements correlated moderate to excellently with CTA (ICCs range 0.52-0.86). No difference in SNR was observed between IR and SR. High quality PV CE-MRA is possible in under 4 min using iNAV bolus timing/motion correction and VD-CASPR.

Magnetic resonance elastography resolving all gross anatomical segments of the kidney during controlled hydration.

Introduction: Magnetic resonance elastography (MRE) is a non-invasive method to quantify biomechanical properties of human tissues. It has potential in diagnosis and monitoring of kidney disease, if established in clinical practice. The interplay of flow and volume changes in renal vessels, tubule, urinary collection system and interstitium is complex, but physiological ranges of in vivo viscoelastic properties during fasting and hydration have never been investigated in all gross anatomical segments simultaneously. Method: Ten healthy volunteers underwent two imaging sessions, one following a 12-hour fasting period and the second after a drinking challenge of >10 mL per kg body weight (60-75 min before the second examination). High-resolution renal MRE was performed using a novel driver with rotating eccentric mass placed at the posterior-lateral wall to couple waves (50 Hz) to the kidney. The biomechanical parameters, shear wave speed (cs in m/s), storage modulus (Gd in kPa), loss modulus (Gl in kPa), phase angle (Υ=2πatanGlGd) and attenuation (α in 1/mm) were derived. Accurate separation of gross anatomical segments was applied in post-processing (whole kidney, cortex, medulla, sinus, vessel). Results: High-quality shear waves coupled into all gross anatomical segments of the kidney (mean shear wave displacement: 163 ± 47 µm, mean contamination of second upper harmonics <23%, curl/divergence: 4.3 ± 0.8). Regardless of the hydration state, median Gd of the cortex and medulla (0.68 ± 0.11 kPa) was significantly higher than that of the sinus and vessels (0.48 ± 0.06 kPa), and consistently, significant differences were found in cs, Υ, and Gl (all p < 0.001). The viscoelastic parameters of cortex and medulla were not significantly different. After hydration sinus exhibited a small but significant reduction in median Gd by -0.02 ± 0.04 kPa (p = 0.01), and, consequently, the cortico-sinusoidal-difference in Gd increased by 0.04 ± 0.07 kPa (p = 0.05). Only upon hydration, the attenuation in vessels became lower (0.084 ± 0.013 1/mm) and differed significantly from the whole kidney (0.095 ± 0.007 1/mm, p = 0.01). Conclusion: High-resolution renal MRE with an innovative driver and well-defined 3D segmentation can resolve all renal segments, especially when including the sinus in the analysis. Even after a prolonged hydration period the approach is sensitive to small hydration-related changes in the sinus and in the cortico-sinusoidal-difference.

Non-rigid motion-compensated 3D whole-heart T2 mapping in a hybrid 3T PET-MR system.

PURPOSE: Simultaneous PET-MRI improves inflammatory cardiac disease diagnosis. However, challenges persist in respiratory motion and mis-registration between free-breathing 3D PET and 2D breath-held MR images. We propose a free-breathing non-rigid motion-compensated 3D T2 -mapping sequence enabling whole-heart myocardial tissue characterization in a hybrid 3T PET-MR system and provides non-rigid respiratory motion fields to correct also simultaneously acquired PET data. METHODS: Free-breathing 3D whole-heart T2 -mapping was implemented on a hybrid 3T PET-MRI system. Three datasets were acquired with different T2 -preparation modules (0, 28, 55 ms) using 3-fold undersampled variable-density Cartesian trajectory. Respiratory motion was estimated via virtual 3D image navigators, enabling multi-contrast non-rigid motion-corrected MR reconstruction. T2 -maps were computed using dictionary-matching. Approach was tested in phantom, 8 healthy subjects, 14 MR only and 2 PET-MR patients with suspected cardiac disease and compared with spin echo reference (phantom) and clinical 2D T2 -mapping (in-vivo). RESULTS: Phantom results show a high correlation (R2 = 0.996) between proposed approach and gold standard 2D T2 mapping. In-vivo 3D T2 -mapping average values in healthy subjects (39.0 ± 1.4 ms) and patients (healthy tissue) (39.1 ± 1.4 ms) agree with conventional 2D T2 -mapping (healthy = 38.6 ± 1.2 ms, patients = 40.3 ± 1.7 ms). Bland-Altman analysis reveals bias of 1.8 ms and 95% limits of agreement (LOA) of -2.4-6 ms for healthy subjects, and bias of 1.3 ms and 95% LOA of -1.9 to 4.6 ms for patients. CONCLUSION: Validated efficient 3D whole-heart T2 -mapping at hybrid 3T PET-MRI provides myocardial inflammation characterization and non-rigid respiratory motion fields for simultaneous PET data correction. Comparable T2 values were achieved with both 3D and 2D methods. Improved image quality was observed in the PET images after MR-based motion correction.

Referenceless Nyquist ghost correction outperforms standard navigator-based method and improves efficiency of in vivo diffusion tensor cardiovascular magnetic resonance.

PURPOSE: The study aims to assess the potential of referenceless methods of EPI ghost correction to accelerate the acquisition of in vivo diffusion tensor cardiovascular magnetic resonance (DT-CMR) data using both computational simulations and data from in vivo experiments. METHODS: Three referenceless EPI ghost correction methods were evaluated on mid-ventricular short axis DT-CMR spin echo and STEAM datasets from 20 healthy subjects at 3T. The reduced field of view excitation technique was used to automatically quantify the Nyquist ghosts, and DT-CMR images were fit to a linear ghost model for correction. RESULTS: Numerical simulation showed the singular value decomposition (SVD) method is the least sensitive to noise, followed by Ghost/Object method and entropy-based method. In vivo experiments showed significant ghost reduction for all correction methods, with referenceless methods outperforming navigator methods for both spin echo and STEAM sequences at b = 32, 150, 450, and 600 smm - 2 $$ {\mathrm{smm}}^{-2} $$ . It is worth noting that as the strength of the diffusion encoding increases, the performance gap between the referenceless method and the navigator-based method diminishes. CONCLUSION: Referenceless ghost correction effectively reduces Nyquist ghost in DT-CMR data, showing promise for enhancing the accuracy and efficiency of measurements in clinical practice without the need for any additional reference scans.

Characterization of Quantitative Susceptibility Mapping in the Left Ventricle of the Myocardium.

BACKGROUND: Myocardial quantitative susceptibility mapping (QSM) may offer better specificity to iron than conventional T2* imaging in the assessment of cardiac diseases, including intra-myocardial haemorrhage. However, the precision and repeatability of cardiac QSM has not yet been characterized. The aim of this study is to characterize these key metrics in a healthy volunteer cohort and show the feasibility of the method in patients. METHODS: Free breathing respiratory-navigated multi-echo 3D gradient echo (GRE) images were acquired, from which QSM maps were reconstructed using the MEDI toolbox. This technique was first evaluated in a susceptibility phantom containing tubes with known concentrations of gadolinium. In vivo characterization of myocardial QSM were then performed in a cohort of 10 healthy volunteers where each subject was scanned twice. Mean segment susceptibility, precision (standard deviation of voxel magnetic susceptibilities within one segment), and repeatability (absolute difference in segment mean susceptibility between repeats) of QSM were calculated for each AHA myocardial segment. Finally, the feasibility of the method was shown in ten patients, including four with haemorrhagic infarcts. RESULTS: The phantom experiment showed a strong linear relationship between measured and predicted susceptibility shifts (R2 > 0.99). For the healthy volunteer cohort, AHA segment analysis showed the mean segment susceptibility was 0.00 ± 0.02 ppm, the mean precision was 0.05 ± 0.04 ppm, and the mean repeatability was 0.02 ± 0.02 ppm. Cardiac QSM was successfully performed in all patients. Focal iron deposits were successfully visualized in the patients with haemorrhagic myocardial infarctions. CONCLUSION: The precision and repeatability of cardiac QSM were successfully characterized in phantom and in vivo experiments. The feasibility of the technique was also successfully demonstrated in patients. While challenges still remain, further clinical evaluation of the technique is now warranted. TRIAL REGISTRATION: This work does not report on a health care intervention.

Feasibility and image quality of bright-blood and black-blood phase-sensitive inversion recovery (BOOST) sequence in clinical practice using for left atrial visualization in patients with atrial fibrillation.

OBJECTIVES: Visualizing left atrial anatomy including the pulmonary veins (PVs) is important for planning the procedure of pulmonary vein isolation with ablation in patients with atrial fibrillation (AF). The aims of our study are to investigate the feasibility of the 3D whole-heart bright-blood and black-blood phase-sensitive (BOOST) inversion recovery sequence in patients with AF scheduled for ablation or electro-cardioversion, and to analyze the correlation between image quality and heart rate and rhythm of patients. METHODS: BOOST was performed for assessing PVs both with T2 preparation pre-pulse (T2prep) and magnetization transfer preparation (MTC) in 45 patients with paroxysmal or permanent AF scheduled for ablation or electro-cardioversion. Image quality analyses were performed by two independent observers. Qualitative assessment was made using the Likert scale; for quantitative analysis, signal to noise ratios (SNR) and contrast to noise ratios (CNR) were calculated for each PV. Heart rate and rhythm were analyzed based on standard 12-lead ECGs. RESULTS: All MTC-BOOST acquisitions achieved diagnostic quality in the PVs, while a significant proportion of T2prep-BOOST images were not suitable for assessing PVs. SNR and CNR values of the MTC-BOOST bright-blood images were higher if patients had sinus rhythm. We found a significant or nearly significant negative correlation between heart rate and the SNR and CNR values of MTC-BOOST bright-blood images. CONCLUSION: 3D whole-heart MTC-BOOST bright-blood imaging is suitable for visualizing the PVs in patients with AF, producing diagnostic image quality in 100% of cases. However, image quality was influenced by heart rate and rhythm. CLINICAL RELEVANCE STATEMENT: The novel 3D whole-heart BOOST CMR sequence needs no contrast administration and is performed during free-breathing; therefore, it is easy to use for a wide range of patients and is suitable for visualizing the PVs in patients with AF. KEY POINTS: • The applicability of the novel 3D whole-heart bright-blood and black-blood phase-sensitive sequence to pulmonary vein imaging in clinical practice is unknown. • Magnetization transfer-bright-blood and black-blood phase-sensitive imaging is suitable for visualizing the pulmonary veins in patients with atrial fibrillation with excellent or good image quality. • Bright-blood and black-blood phase-sensitive cardiac magnetic resonance sequence is easy to use for a wide range of patients as it needs no contrast administration and is performed during free-breathing.

Real-time automatic image-based slice tracking of gadolinium-filled balloon wedge catheter during MR-guided cardiac catheterization: A proof-of-concept study.

PURPOSE: MR-guided cardiac catheterization procedures currently use passive tracking approaches to follow a gadolinium-filled catheter balloon during catheter navigation. This requires frequent manual tracking and repositioning of the imaging slice during navigation. In this study, a novel framework for automatic real-time catheter tracking during MR-guided cardiac catheterization is presented. METHODS: The proposed framework includes two imaging modes (Calibration and Runtime). The sequence starts in Calibration mode, in which the 3D catheter coordinates are determined using a stack of 10-20 contiguous saturated slices combined with real-time image processing. The sequence then automatically switches to Runtime mode, where three contiguous slices (acquired with partial saturation), initially centered on the catheter balloon using the Calibration feedback, are acquired continuously. The 3D catheter balloon coordinates are estimated in real time from each Runtime slice stack using image processing. Each Runtime stack is repositioned to maintain the catheter balloon in the central slice based on the prior Runtime feedback. The sequence switches back to Calibration mode if the catheter is not detected. This framework was evaluated in a heart phantom and 3 patients undergoing MR-guided cardiac catheterization. Catheter detection accuracy and rate of catheter visibility were evaluated. RESULTS: The automatic detection accuracy for the catheter balloon during the Calibration/Runtime mode was 100%/95% in phantom and 100%/97 ± 3% in patients. During Runtime, the catheter was visible in 82% and 98 ± 2% of the real-time measurements in the phantom and patients, respectively. CONCLUSION: The proposed framework enabled real-time continuous automatic tracking of a gadolinium-filled catheter balloon during MR-guided cardiac catheterization.

Combining generative modelling and semi-supervised domain adaptation for whole heart cardiovascular magnetic resonance angiography segmentation.

BACKGROUND: Quantification of three-dimensional (3D) cardiac anatomy is important for the evaluation of cardiovascular diseases. Changes in anatomy are indicative of remodeling processes as the heart tissue adapts to disease. Although robust segmentation methods exist for computed tomography angiography (CTA), few methods exist for whole-heart cardiovascular magnetic resonance angiograms (CMRA) which are more challenging due to variable contrast, lower signal to noise ratio and a limited amount of labeled data. METHODS: Two state-of-the-art unsupervised generative deep learning domain adaptation architectures, generative adversarial networks and variational auto-encoders, were applied to 3D whole heart segmentation of both conventional (n = 20) and high-resolution (n = 45) CMRA (target) images, given segmented CTA (source) images for training. An additional supervised loss function was implemented to improve performance given 10%, 20% and 30% segmented CMRA cases. A fully supervised nn-UNet trained on the given CMRA segmentations was used as the benchmark. RESULTS: The addition of a small number of segmented CMRA training cases substantially improved performance in both generative architectures in both standard and high-resolution datasets. Compared with the nn-UNet benchmark, the generative methods showed substantially better performance in the case of limited labelled cases. On the standard CMRA dataset, an average 12% (adversarial method) and 10% (variational method) improvement in Dice score was obtained. CONCLUSIONS: Unsupervised domain-adaptation methods for CMRA segmentation can be boosted by the addition of a small number of supervised target training cases. When only few labelled cases are available, semi-supervised generative modelling is superior to supervised methods.

Automated detection of cardiac rest period for trigger delay calculation for image-based navigator coronary magnetic resonance angiography.

BACKGROUND: Coronary magnetic resonance angiography (coronary MRA) is increasingly being considered as a clinically viable method to investigate coronary artery disease (CAD). Accurate determination of the trigger delay to place the acquisition window within the quiescent part of the cardiac cycle is critical for coronary MRA in order to reduce cardiac motion. This is currently reliant on operator-led decision making, which can negatively affect consistency of scan acquisition. Recently developed deep learning (DL) derived software may overcome these issues by automation of cardiac rest period detection. METHODS: Thirty individuals (female, n = 10) were investigated using a 0.9 mm isotropic image-navigator (iNAV)-based motion-corrected coronary MRA sequence. Each individual was scanned three times utilising different strategies for determination of the optimal trigger delay: (1) the DL software, (2) an experienced operator decision, and (3) a previously utilised formula for determining the trigger delay. Methodologies were compared using custom-made analysis software to assess visible coronary vessel length and coronary vessel sharpness for the entire vessel length and the first 4 cm of each vessel. RESULTS: There was no difference in image quality between any of the methodologies for determination of the optimal trigger delay, as assessed by visible coronary vessel length, coronary vessel sharpness for each entire vessel and vessel sharpness for the first 4 cm of the left mainstem, left anterior descending or right coronary arteries. However, vessel length of the left circumflex was slightly greater using the formula method. The time taken to calculate the trigger delay was significantly lower for the DL-method as compared to the operator-led approach (106 ± 38.0 s vs 168 ± 39.2 s, p < 0.01, 95% CI of difference 25.5-98.1 s). CONCLUSIONS: Deep learning-derived automated software can effectively and efficiently determine the optimal trigger delay for acquisition of coronary MRA and thus may simplify workflow and improve reproducibility.

Feasibility of cardiac MR thermometry at 0.55†T.

Radiofrequency catheter ablation is an established treatment strategy for ventricular tachycardia, but remains associated with a low success rate. MR guidance of ventricular tachycardia shows promises to improve the success rate of these procedures, especially due to its potential to provide real-time information on lesion formation using cardiac MR thermometry. Modern low field MRI scanners (<1 T) are of major interest for MR-guided ablations as the potential benefits include lower costs, increased patient access and device compatibility through reduced device-induced imaging artefacts and safety constraints. However, the feasibility of cardiac MR thermometry at low field remains unknown. In this study, we demonstrate the feasibility of cardiac MR thermometry at 0.55 T and characterized its in vivo stability (i.e., precision) using state-of-the-art techniques based on the proton resonance frequency shift method. Nine healthy volunteers were scanned using a cardiac MR thermometry protocol based on single-shot EPI imaging (3 slices in the left ventricle, 150 dynamics, TE = 41 ms). The reconstruction pipeline included image registration to align all the images, multi-baseline approach (look-up-table length = 30) to correct for respiration-induced phase variations, and temporal filtering to reduce noise in temperature maps. The stability of thermometry was defined as the pixel-wise standard deviation of temperature changes over time. Cardiac MR thermometry was successfully acquired in all subjects and the stability averaged across all subjects was 1.8 ± 1.0°C. Without multi-baseline correction, the overall stability was 2.8 ± 1.6°C. In conclusion, cardiac MR thermometry is feasible at 0.55 T and further studies on MR-guided catheter ablations at low field are warranted.

Automatic image-based tracking of gadolinium-filled balloon wedge catheters for MRI-guided cardiac catheterization using deep learning.

Introduction: Magnetic Resonance Imaging (MRI) is a promising alternative to standard x-ray fluoroscopy for the guidance of cardiac catheterization procedures as it enables soft tissue visualization, avoids ionizing radiation and provides improved hemodynamic data. MRI-guided cardiac catheterization procedures currently require frequent manual tracking of the imaging plane during navigation to follow the tip of a gadolinium-filled balloon wedge catheter, which unnecessarily prolongs and complicates the procedures. Therefore, real-time automatic image-based detection of the catheter balloon has the potential to improve catheter visualization and navigation through automatic slice tracking. Methods: In this study, an automatic, parameter-free, deep-learning-based post-processing pipeline was developed for real-time detection of the catheter balloon. A U-Net architecture with a ResNet-34 encoder was trained on semi-artificial images for the segmentation of the catheter balloon. Post-processing steps were implemented to guarantee a unique estimate of the catheter tip coordinates. This approach was evaluated retrospectively in 7 patients (6M and 1F, age = 7 ± 5 year) who underwent an MRI-guided right heart catheterization procedure with all images acquired in an orientation unseen during training. Results: The overall accuracy, specificity and sensitivity of the proposed catheter tracking strategy over all 7 patients were 98.4 ± 2.0%, 99.9 ± 0.2% and 95.4 ± 5.5%, respectively. The computation time of the deep-learning-based segmentation step was ∼10 ms/image, indicating its compatibility with real-time constraints. Conclusion: Deep-learning-based catheter balloon tracking is feasible, accurate, parameter-free, and compatible with real-time conditions. Online integration of the technique and its evaluation in a larger patient cohort are now warranted to determine its benefit during MRI-guided cardiac catheterization.

Use of a new non-contrast-enhanced BOOST cardiac MR sequence before electrical cardioversion or ablation of atrial fibrillation-a pilot study.

Introduction: Left atrial appendage (LAA) thrombus is the most common source of embolization in atrial fibrillation (AF). Transesophageal echocardiography (TEE) is the gold standard method for LAA thrombus exclusion. Our pilot study aimed to compare the efficacy of a new non-contrast-enhanced cardiac magnetic resonance (CMR) sequence (BOOST) with TEE for the detection of LAA thrombus and to evaluate the usefulness of BOOST images for planning radiofrequency catheter ablation (RFCA) compared with left atrial (LA) contrast-enhanced computed tomography (CT). We also attempted to assess the patients' subjective experiences with TEE and CMR. Methods: Patients with AF undergoing either electrical cardioversion or RFCA were enrolled. Participants underwent pre-procedural TEE and CMR scans to evaluate LAA thrombus status and pulmonary vein anatomy. Patient experiences with TEE and CMR were assessed using a questionnaire developed by our team. Some patients scheduled for RFCA also had pre-procedural LA contrast-enhanced CT. In such cases, the operating physician was asked to subjectively define the quality of the CT and CMR scan on a scale of 1-10 (1 = worst, 10 = best) and comment on CMR's usefulness in RFCA planning. Results: Seventy-one patients were enrolled. In 94.4%, both TEE and CMR excluded, and in 1 patient, both modalities reported the presence of LAA thrombus. In 1 patient, TEE was inconclusive, but CMR excluded LAA thrombus. In 2 patients, CMR could not exclude the presence of thrombus, but in 1 of those cases, TEE was also indecisive. During TEE, 67%, during CMR, only 1.9% of patients reported pain (p < 0.0001), and 89% would prefer CMR in case of a repeat examination. The quality of the left atrial contrast-enhanced CT scans was better compared with the image quality of the CMR BOOST sequence [8 (7-9) vs. 6 (5-7), p < 0.0001]. Still, the CMR images were useful for procedural planning in 91% of cases. Conclusion: The new CMR BOOST sequence provides appropriate image quality for ablation planning. The sequence might be useful for excluding larger LAA thrombi; however, its accuracy in detecting smaller thrombi is limited. Most patients preferred CMR over TEE in this indication.

Real-time fetal brain tracking for functional fetal MRI.

PURPOSE: To improve motion robustness of functional fetal MRI scans by developing an intrinsic real-time motion correction method. MRI provides an ideal tool to characterize fetal brain development and growth. It is, however, a relatively slow imaging technique and therefore extremely susceptible to subject motion, particularly in functional MRI experiments acquiring multiple Echo-Planar-Imaging-based repetitions, for example, diffusion MRI or blood-oxygen-level-dependency MRI. METHODS: A 3D UNet was trained on 125 fetal datasets to track the fetal brain position in each repetition of the scan in real time. This tracking, inserted into a Gadgetron pipeline on a clinical scanner, allows updating the position of the field of view in a modified echo-planar imaging sequence. The method was evaluated in real-time in controlled-motion phantom experiments and ten fetal MR studies (17 + 4-34 + 3 gestational weeks) at 3T. The localization network was additionally tested retrospectively on 29 low-field (0.55T) datasets. RESULTS: Our method achieved real-time fetal head tracking and prospective correction of the acquisition geometry. Localization performance achieved Dice scores of 84.4% and 82.3%, respectively for both the unseen 1.5T/3T and 0.55T fetal data, with values higher for cephalic fetuses and increasing with gestational age. CONCLUSIONS: Our technique was able to follow the fetal brain even for fetuses under 18 weeks GA in real-time at 3T and was successfully applied "offline" to new cohorts on 0.55T. Next, it will be deployed to other modalities such as fetal diffusion MRI and to cohorts of pregnant participants diagnosed with pregnancy complications, for example, pre-eclampsia and congenital heart disease.

Pre-training via Transfer Learning and Pretext Learning a Convolutional Neural Network for Automated Assessments of Clinical PET Image Quality.

Positron emission tomography (PET) using a fraction of the usual injected dose would reduce the amount of radioligand needed, as well as the radiation dose to patients and staff, but would compromise reconstructed image quality. For performing the same clinical tasks with such images, a clinical (rather than numerical) image quality assessment is essential. This process can be automated with convolutional neural networks (CNNs). However, the scarcity of clinical quality readings is a challenge. We hypothesise that exploiting easily available quantitative information in pretext learning tasks or using established pre-trained networks could improve CNN performance for predicting clinical assessments with limited data. CNNs were pre-trained to predict injected dose from image patches extracted from eight real patient datasets, reconstructed using between 0.5%-100% of the available data. Transfer learning with seven different patients was used to predict three clinically-scored quality metrics ranging from 0-3: global quality rating, pattern recognition and diagnostic confidence. This was compared to pre-training via a VGG16 network at varying pre-training levels. Pre-training improved test performance for this task: the mean absolute error of 0.53 (compared to 0.87 without pre-training), was within clinical scoring uncertainty. Future work may include using the CNN for novel reconstruction methods performance assessment.

Free-breathing, Contrast Agent-free Whole-Heart MTC-BOOST Imaging: Single-Center Validation Study in Adult Congenital Heart Disease.

Purpose: To assess the clinical performance of the three-dimensional, free-breathing, Magnetization Transfer Contrast Bright-and-black blOOd phase-SensiTive (MTC-BOOST) sequence in adult congenital heart disease (ACHD). Materials and Methods: In this prospective study, participants with ACHD undergoing cardiac MRI between July 2020 and March 2021 were scanned with the clinical T2-prepared balanced steady-state free precession sequence and proposed MTC-BOOST sequence. Four cardiologists scored their diagnostic confidence on a four-point Likert scale for sequential segmental analysis on images acquired with each sequence. Scan times and diagnostic confidence were compared using the Mann-Whitney test. Coaxial vascular dimensions at three anatomic landmarks were measured, and agreement between the research sequence and the corresponding clinical sequence was assessed with Bland-Altman analysis. Results: The study included 120 participants (mean age, 33 years ± 13 [SD]; 65 men). The mean acquisition time of the MTC-BOOST sequence was significantly lower compared with that of the conventional clinical sequence (9 minutes ± 2 vs 14 minutes ± 5; P < .001). Diagnostic confidence was higher for the MTC-BOOST sequence compared with the clinical sequence (mean, 3.9 ± 0.3 vs 3.4 ± 0.7; P < .001). Narrow limits of agreement and mean bias less than 0.08 cm were found between the research and clinical vascular measurements. Conclusion: The MTC-BOOST sequence provided efficient, high-quality, and contrast agent-free three-dimensional whole-heart imaging in ACHD, with shorter, more predictable acquisition time and improved diagnostic confidence compared with the reference standard clinical sequence.Keywords: MR Angiography, Cardiac Supplemental material is available for this article. Published under a CC BY 4.0 license.

Automated segmentation of long and short axis DENSE cardiovascular magnetic resonance for myocardial strain analysis using spatio-temporal convolutional neural networks.

BACKGROUND: Cine Displacement Encoding with Stimulated Echoes (DENSE) facilitates the quantification of myocardial deformation, by encoding tissue displacements in the cardiovascular magnetic resonance (CMR) image phase, from which myocardial strain can be estimated with high accuracy and reproducibility. Current methods for analyzing DENSE images still heavily rely on user input, making this process time-consuming and subject to inter-observer variability. The present study sought to develop a spatio-temporal deep learning model for segmentation of the left-ventricular (LV) myocardium, as spatial networks often fail due to contrast-related properties of DENSE images. METHODS: 2D + time nnU-Net-based models have been trained to segment the LV myocardium from DENSE magnitude data in short- and long-axis images. A dataset of 360 short-axis and 124 long-axis slices was used to train the networks, from a combination of healthy subjects and patients with various conditions (hypertrophic and dilated cardiomyopathy, myocardial infarction, myocarditis). Segmentation performance was evaluated using ground-truth manual labels, and a strain analysis using conventional methods was performed to assess strain agreement with manual segmentation. Additional validation was performed using an externally acquired dataset to compare the inter- and intra-scanner reproducibility with respect to conventional methods. RESULTS: Spatio-temporal models gave consistent segmentation performance throughout the cine sequence, while 2D architectures often failed to segment end-diastolic frames due to the limited blood-to-myocardium contrast. Our models achieved a DICE score of 0.83 ± 0.05 and a Hausdorff distance of 4.0 ± 1.1 mm for short-axis segmentation, and 0.82 ± 0.03 and 7.9 ± 3.9 mm respectively for long-axis segmentations. Strain measurements obtained from automatically estimated myocardial contours showed good to excellent agreement with manual pipelines, and remained within the limits of inter-user variability estimated in previous studies. CONCLUSION: Spatio-temporal deep learning shows increased robustness for the segmentation of cine DENSE images. It provides excellent agreement with manual segmentation for strain extraction. Deep learning will facilitate the analysis of DENSE data, bringing it one step closer to clinical routine.

Retrospective motion correction through multi-average k-space data elimination (REMAKE) for free-breathing cardiac cine imaging.

PURPOSE: To develop a motion-robust reconstruction technique for free-breathing cine imaging with multiple averages. METHOD: Retrospective motion correction through multiple average k-space data elimination (REMAKE) was developed using iterative removal of k-space segments (from individual k-space samples) that contribute most to motion corruption while combining any remaining segments across multiple signal averages. A variant of REMAKE, termed REMAKE+, was developed to address any losses in SNR due to k-space information removal. With REMAKE+, multiple reconstructions using different initial conditions were performed, co-registered, and averaged. Both techniques were validated against clinical "standard" signal averaging reconstruction in a static phantom (with simulated motion) and 15 patients undergoing free-breathing cine imaging with multiple averages. Quantitative analysis of myocardial sharpness, blood/myocardial SNR, myocardial-blood contrast-to-noise ratio (CNR), as well as subjective assessment of image quality and rate of diagnostic quality images were performed. RESULTS: In phantom, motion artifacts using "standard" (RMS error [RMSE]: 2.2 ± 0.5) were substantially reduced using REMAKE/REMAKE+ (RMSE: 1.5 ± 0.4/1.0 ± 0.4, p < 0.01). In patients, REMAKE/REMAKE+ led to higher myocardial sharpness (0.79 ± 0.09/0.79 ± 0.1 vs. 0.74 ± 0.12 for "standard", p = 0.004/0.04), higher image quality (1.8 ± 0.2/1.9 ± 0.2 vs. 1.6 ± 0.4 for "standard", p = 0.02/0.008), and a higher rate of diagnostic quality images (99%/100% vs. 94% for "standard"). Blood/myocardial SNR for "standard" (94 ± 30/33 ± 10) was higher vs. REMAKE (80 ± 25/28 ± 8, p = 0.002/0.005) and tended to be lower vs. REMAKE+ (105 ± 33/36 ± 12, p = 0.02/0.06). Myocardial-blood CNR for "standard" (61 ± 22) was higher vs. REMAKE (53 ± 19, p = 0.003) and lower vs. REMAKE+ (69 ± 24, p = 0.007). CONCLUSIONS: Compared to "standard" signal averaging reconstruction, REMAKE and REMAKE+ provide improved myocardial sharpness, image quality, and rate of diagnostic quality images.

Simultaneous Highly Efficient Contrast-Free Lumen and Vessel Wall MR Imaging for Anatomical Assessment of Aortic Disease.

BACKGROUND: Bright-blood lumen and black-blood vessel wall imaging are required for the comprehensive assessment of aortic disease. These images are usually acquired separately, resulting in long examinations and potential misregistration between images. PURPOSE: To characterize the performance of an accelerated and respiratory motion-compensated three-dimensional (3D) cardiac MRI technique for simultaneous contrast-free aortic lumen and vessel wall imaging with an interleaved T2 and inversion recovery prepared sequence (iT2Prep-BOOST). STUDY TYPE: Prospective. POPULATION: A total of 30 consecutive patients with aortopathy referred for a clinically indicated cardiac MRI examination (9 females, mean age ± standard deviation: 32 ± 12 years). FIELD STRENGTH/SEQUENCE: 1.5-T; bright-blood MR angiography (diaphragmatic navigator-gated T2-prepared 3D balanced steady-state free precession [bSSFP], T2Prep-bSSFP), breath-held black-blood two-dimensional (2D) half acquisition single-shot turbo spin echo (HASTE), and 3D bSSFP iT2Prep-BOOST. ASSESSMENT: iT2Prep-BOOST bright-blood images were compared to T2prep-bSSFP images in terms of aortic vessel dimensions, lumen-to-myocardium contrast ratio (CR), and image quality (diagnostic confidence, vessel sharpness and presence of artifacts, assessed by three cardiologists on a 4-point scale, 1: nondiagnostic to 4: excellent). The iT2Prep-BOOST black-blood images were compared to 2D HASTE images for quantification of wall thickness. A visual comparison between computed tomography (CT) and iT2Prep-BOOST was performed in a patient with chronic aortic dissection. STATISTICAL TESTS: Paired t-tests, Wilcoxon signed-rank tests, intraclass correlation coefficient (ICC), Bland-Altman analysis. A P value < 0.05 was considered statistically significant. RESULTS: Bright-blood iT2Prep-BOOST resulted in significantly improved image quality (mean ± standard deviation 3.8 ± 0.5 vs. 3.3 ± 0.8) and CR (2.9 ± 0.8 vs. 1.8 ± 0.5) compared with T2Prep-bSSFP, with a shorter scan time (7.8 ± 1.7 minutes vs. 12.9 ± 3.4 minutes) while providing a complementary 3D black-blood image. Aortic lumen diameter and vessel wall thickness measurements in bright-blood and black-blood images were in good agreement with T2Prep-bSSFP and HASTE images (<0.02 cm and <0.005 cm bias, respectively) and good intrareader (ICC > 0.96) and interreader (ICC > 0.94) agreement was observed for all measurements. DATA CONCLUSION: iT2Prep-BOOST might enable time-efficient simultaneous bright- and black-blood aortic imaging, with improved image quality compared to T2Prep-bSSFP and HASTE imaging, and comparable measurements for aortic wall and lumen dimensions. EVIDENCE LEVEL: 2. TECHNICAL EFFICACY: Stage 2.

Deep learning estimation of three-dimensional left atrial shape from two-chamber and four-chamber cardiac long axis views.

AIMS: Left atrial volume is commonly estimated using the bi-plane area-length method from two-chamber (2CH) and four-chamber (4CH) long axes views. However, this can be inaccurate due to a violation of geometric assumptions. We aimed to develop a deep learning neural network to infer 3D left atrial shape, volume and surface area from 2CH and 4CH views. METHODS AND RESULTS: A 3D UNet was trained and tested using 2CH and 4CH segmentations generated from 3D coronary computed tomography angiography (CCTA) segmentations (n = 1700, with 1400/100/200 cases for training/validating/testing). An independent test dataset from another institution was also evaluated, using cardiac magnetic resonance (CMR) 2CH and 4CH segmentations as input and 3D CCTA segmentations as the ground truth (n = 20). For the 200 test cases generated from CCTA, the network achieved a mean Dice score value of 93.7%, showing excellent 3D shape reconstruction from two views compared with the 3D segmentation Dice of 97.4%. The network also showed significantly lower mean absolute error values of 3.5 mL/4.9 cm2 for LA volume/surface area respectively compared to the area-length method errors of 13.0 mL/34.1 cm2 respectively (P < 0.05 for both). For the independent CMR test set, the network achieved accurate 3D shape estimation (mean Dice score value of 87.4%), and a mean absolute error values of 6.0 mL/5.7 cm2 for left atrial volume/surface area respectively, significantly less than the area-length method errors of 14.2 mL/19.3 cm2 respectively (P < 0.05 for both). CONCLUSIONS: Compared to the bi-plane area-length method, the network showed higher accuracy and robustness for both volume and surface area.