Machine learning for the automatic assessment of aortic rotational flow and wall shear stress from 4D flow cardiac magnetic resonance imaging

Cardiovascular Aging and Risk Assessment: How Multimodality Imaging Can Help

Aging affects the cardiovascular system, and this process may be accelerated in individuals with cardiovascular risk factors. The main vascular changes include arterial wall thickening, calcification, and stiffening, together with aortic dilatation and elongation. With aging, we can observe left ventricular hypertrophy with myocardial fibrosis and left atrial dilatation. These changes may lead to heart failure and atrial fibrillation. Using multimodality imaging, including ultrasound, computed tomography (CT), and magnetic resonance imaging, it is possible to detect these changes. Additionally, multimodality imaging, mainly via CT measurements of coronary artery calcium or ultrasound carotid intima-media thickness, enables advanced cardiovascular risk stratification and helps in decision-making about preventive strategies. The focus of this manuscript is to briefly review cardiovascular changes that occur with aging, as well as to describe how multimodality imaging may be used for the assessment of these changes and risk stratification of asymptomatic individuals.

Assessing Aortic Motion with Automated 3D Cine Balanced SSFP MRI Segmentation

PURPOSE

To apply free-running three-dimensional (3D) cine balanced steady state free precession (bSSFP) CMR framework in combination with AI segmentations to quantify time-resolved aortic displacement, diameter and diameter change.

METHODS

In this prospective study, we implemented a free-running 3D cine bSSFP sequence with scan time of about 4minutes facilitated by pseudo-spiral Cartesian undersampling and compressed-sensing reconstruction. Automated segmentation of all cardiac timeframes was applied through the use of nnU-Net. Dynamic 3D motion maps were created for three repeated scans per volunteer, leading to the detailed quantification of motion, as well as the measurement and change in diameter of the ascending aorta.

RESULTS

A total of 14 adult healthy volunteers (median age, 28 years (IQR: 26.0-31.3), 6 female) were included. Automated segmentation compared to manual segmentation of the aorta test set showed a Dice score of 0.93 ± 0.02. The median (interquartile range) over all volunteers for the largest maximum and mean ascending aorta (AAo) displacement in the first scan was 13.0 (4.4) mm and 5.6 (2.4) mm, respectively. Peak mean diameter in the AAo was 25.9 (2.2) mm and peak mean diameter change was 1.4 (0.5) mm. The maximum individual variability over the three repeated scans of maximum and mean AAo displacement was 3.9 (1.6) mm and 2.2 (0.8) mm, respectively. The maximum individual variability of mean diameter and diameter change were 1.2 (0.5) mm and 0.9 (0.4) mm.

CONCLUSION

A free-running 3D cine bSSFP CMR scan with a scan time of four minutes combined with an automated nnU-net segmentation consistently captured the aorta's cardiac motion-related 4D displacement, diameter, and diameter change.

Impact of valve-sparing aortic root replacement on aortic fluid dynamics and biomechanics in patients with syndromic heritable thoracic aortic disease

OBJECTIVES

Patients with syndromic heritable thoracic aortic diseases (sHTAD) who underwent prophylactic aortic root replacement are at high risk of distal aortic events, but the underlying mechanisms are poorly understood. This prospective, longitudinal study aims to assess the impact of valve-sparing aortic root replacement (VSARR) on aortic fluid dynamics and biomechanics in these patients, and to examine whether they present altered haemodynamics or biomechanics prior to surgery compared to sHTAD patients with no indication for surgery (sHTAD-NSx) and healthy volunteers (HV).

METHODS

Sixteen patients with Marfan or Loeys-Dietz syndrome underwent two 4D flow CMR studies before (sHTAD-preSx) and after VSARR (sHTAD-postSx). Two age, sex and BSA matched cohorts of 40 HV and 16 sHTAD-NSx patients with available 4D flow CMR, were selected for comparison. In-plane rotational flow (IRF), systolic flow reversal ratio (SFRR), wall shear stress (WSS), pulse wave velocity (PWV) and aortic strain were analysed in the ascending (AscAo) and descending aorta (DescAo).

RESULTS

All patients with sHTAD presented altered haemodynamics and increased aortic stiffness (p<0.05) compared to HV, both in the AscAo (median PWV 7.4 in sHTAD-NSx; 6.8 in sHTAD-preSx; 4.9m/s in HV) and DescAo (median PWV 9.1 in sHTAD-NSx; 8.1 in sHTAD-preSx; 6.3m/s in HV). Patients awaiting VSARR had markedly reduced in-plane (median IRF -2.2 vs 10.4 cm2/s in HV, p=0.001), but increased through-plane flow rotation (median SFRR 7.8 vs 3.8% in HV, p=0.002), and decreased WSS (0.36 vs 0.47N/m2 in HV, p=0.004) in the proximal DescAo. After VSARR, proximal DescAo in-plane rotational flow (p=0.010) and circumferential WSS increased (p=0.011), no longer differing from HV, but through-plane rotational flow, axial WSS and stiffness remained altered. Patients in which aortic tortuosity was reduced after surgery showed greater post-surgical increase in IRF compared to those in which tortuosity increased (median IRF increase 18.1 vs 3.3cm²/s, p=0.047). Most AscAo flow alterations were restored to physiological values after VSARR.

CONCLUSIONS

In patients with sHTAD, VSARR partially restores downstream fluid dynamics to physiological levels. However, some flow disturbances and increased stiffness persist in the proximal DescAo. Further longitudinal studies are needed to evaluate whether persistent alterations contribute to post-surgical risk.

Comprehensive sex-specific and age-dependent analysis of 4D-flow MRI assessed aortic blood flow-related parameters in normal subjects using single-vendor MR systems and single-vendor software

BACKGROUND

Aortic blood flow characterization by 4D flow MRI is increasingly performed in aneurysm research. A limited number of studies have established normal values that can aid the recognition of abnormal flow at an early stage. This study aims to establish additional sex-specific and age-dependent reference values for flow-related parameters in a large cohort of healthy adults.

METHODS

212 volunteers were included, and 191 volunteers completed the full study protocol. All underwent 4D flow MRI of the entire aorta. Quantitative values for velocity, vorticity, helicity, as well as total, circumferential, and axial wall shear stress [WSS] were determined for the aortic root [AoR], ascending aorta [AAo], aortic arch [AoA], descending [DAo], suprarenal [SRA], and infrarenal aorta [IRA]. Vorticity and helicity were indexed for segment volume (mL).

RESULTS

The normal values were estimated per sex- and age-group, where significant differences between males (M) and females (F) were found only for specific age groups. More specifically, the following variables were significantly different after applying the false discovery rate correction for multiple testing: 1) velocity in the AAo and DAo in the 60-70 years age group (mean±SD: (M) 47.0 ± 8.2cm/s vs. (F) 38.4 ± 6.9cm/s, p=0.001 and, (M) 55.9 ± 9.9cm/s vs. (F) 46.5 ± 5.5cm/s, p=0.002), 2) normalized vorticity in AoR in the 50-59 years age group ((M) 27539 ± 5042s-1mL-1 vs. (F) 30849 ± 7285s-1mL-1, p=0.002), 3) axial WSS in the Aao in the 18-29 age group ((M) 1098 ± 203 mPa vs. (F) 921 ± 121 mPa, p=0.002). Good to strong negative correlations with age were seen for almost all variables, in different segments, and for both sexes.

CONCLUSION

This study describes reference values for aortic flow-related parameters as acquired by 4D flow MRI. We observed limited differences between males and females. A negative relationship with age was seen for almost all flow-related parameters and segments.

Evaluating a Phase-Specific Approach to Aortic Flow: A 4D Flow MRI Study

BACKGROUND

Aortic flow parameters can be quantified using 4D flow MRI. However, data are sparse on how different methods of analysis influence these parameters and how these parameters evolve during systole.

PURPOSE

To assess multiphase segmentations and multiphase quantification of flow-related parameters in aortic 4D flow MRI.

STUDY TYPE

Prospective.

POPULATION

40 healthy volunteers (50% male, 28.9 ± 5.0 years) and 10 patients with thoracic aortic aneurysm (80% male, 54 ± 8 years).

FIELD STRENGTH/SEQUENCE

4D flow MRI with a velocity encoded turbo field echo sequence at 3 T.

ASSESSMENT

Phase-specific segmentations were obtained for the aortic root and the ascending aorta. The whole aorta was segmented in peak systole. In all aortic segments, time to peak (TTP; for flow velocity, vorticity, helicity, kinetic energy, and viscous energy loss) and peak and time-averaged values (for velocity and vorticity) were calculated.

STATISTICAL TESTS

Static vs. phase-specific models were assessed using Bland-Altman plots. Other analyses were performed using phase-specific segmentations for aortic root and ascending aorta. TTP for all parameters was compared to TTP of flow rate using paired t-tests. Time-averaged and peak values were assessed using Pearson correlation coefficient. P < 0.05 was considered statistically significant.

RESULTS

In the combined group, velocity in static vs. phase-specific segmentations differed by 0.8 cm/sec for the aortic root, and 0.1 cm/sec (P = 0.214) for the ascending aorta. Vorticity differed by 167 sec-1  mL-1 (P = 0.468) for the aortic root, and by 59 sec-1  mL-1 (P = 0.481) for the ascending aorta. Vorticity, helicity, and energy loss in the ascending aorta, aortic arch, and descending aorta peaked significantly later than flow rate. Time-averaged velocity and vorticity values correlated significantly in all segments.

DATA CONCLUSION

Static 4D flow MRI segmentation yields comparable results as multiphase segmentation for flow-related parameters, eliminating the need for time-consuming multiple segmentations. However, multiphase quantification is necessary for assessing peak values of aortic flow-related parameters.

LEVEL OF EVIDENCE

2 TECHNICAL EFFICACY STAGE: 3.

Deep Learning-Based Analysis of Aortic Morphology From Three-Dimensional MRI

BACKGROUND

Quantification of aortic morphology plays an important role in the evaluation and follow-up assessment of patients with aortic diseases, but often requires labor-intensive and operator-dependent measurements. Automatic solutions would help enhance their quality and reproducibility.

PURPOSE

To design a deep learning (DL)-based automated approach for aortic landmarks and lumen detection derived from three-dimensional (3D) MRI.

STUDY TYPE

Retrospective.

POPULATION

Three hundred ninety-one individuals (female: 47%, age = 51.9 ± 18.4) from three sites, including healthy subjects and patients (hypertension, aortic dilation, Turner syndrome), randomly divided into training/validation/test datasets (N = 236/77/78). Twenty-five subjects were randomly selected and analyzed by three operators with different levels of expertise.

FIELD STRENGTH/SEQUENCE

1.5-T and 3-T, 3D spoiled gradient-recalled or steady-state free precession sequences.

ASSESSMENT

Reinforcement learning and a two-stage network trained using reference landmarks and segmentation from an existing semi-automatic software were used for aortic landmark detection and segmentation from sinotubular junction to coeliac trunk. Aortic segments were defined using the detected landmarks while the aortic centerline was extracted from the segmentation and morphological indices (length, aortic diameter, and volume) were computed for both the reference and the proposed segmentations.

STATISTICAL TESTS

Segmentation: Dice similarity coefficient (DSC), Hausdorff distance (HD), average symmetrical surface distance (ASSD); landmark detection: Euclidian distance (ED); model robustness: Spearman correlation, Bland-Altman analysis, Kruskal-Wallis test for comparisons between reference and DL-derived aortic indices; inter-observer study: Williams index (WI). A WI 95% confidence interval (CI) lower bound >1 indicates that the method is within the inter-observer variability. A P-value <0.05 was considered statistically significant.

RESULTS

DSC was 0.90 ± 0.05, HD was 12.11 ± 7.79 mm, and ASSD was 1.07 ± 0.63 mm. ED was 5.0 ± 6.1 mm. A good agreement was found between all DL-derived and reference aortic indices (r >0.95, mean bias <7%). Our segmentation and landmark detection performances were within the inter-observer variability except the sinotubular junction landmark (CI = 0.96;1.04).

DATA CONCLUSION

A DL-based aortic segmentation and anatomical landmark detection approach was developed and applied to 3D MRI data for achieve aortic morphology evaluation.

EVIDENCE LEVEL

3 TECHNICAL EFFICACY: Stage 2.

Three-dimensional aortic geometry mapping via registration of non-gated contrast-enhanced or gated and respiratory-navigated MR angiographies

BACKGROUND

The measurement of aortic dimensions and their evolution are key in the management of patients with aortic diseases. Manual assessment, the current guideline-recommended method and clinical standard, is subjective, poorly reproducible, and time-consuming, limiting the capacity to track aortic growth in everyday practice. Aortic geometry mapping (AGM) via image registration of serial computed tomography angiograms outperforms manual assessment, providing accurate and reproducible 3D maps of aortic diameter and growth rate. This observational study aimed to evaluate the accuracy and reproducibility of AGM on non-gated contrast-enhanced (CE-) and cardiac- and respiratory-gated (GN-) magnetic resonance angiographies (MRA).

METHODS

Patients with thoracic aortic disease followed with serial CE-MRA (n = 30) or GN-MRA (n = 15) acquired at least 1 year apart were retrospectively and consecutively identified. Two independent observers measured aortic diameters and growth rates (GR) manually at several thoracic aorta reference levels and with AGM. Agreement between manual and AGM measurements and their inter-observer reproducibility were compared. Reproducibility for aortic diameter and GR maps assessed with AGM was obtained.

RESULTS

Mean follow-up was 3.8 ± 2.3 years for CE- and 2.7 ± 1.6 years for GN-MRA. AGM was feasible in the 93% of CE-MRA pairs and in the 100% of GN-MRA pairs. Manual and AGM diameters showed excellent agreement and inter-observer reproducibility (ICC>0.9) at all anatomical levels. Agreement between manual and AGM GR was more limited, both in the aortic root by GN-MRA (ICC=0.47) and in the thoracic aorta, where higher accuracy was obtained with GN- than with CE-MRA (ICC=0.55 vs 0.43). The inter-observer reproducibility of GR by AGM was superior compared to manual assessment, both with CE- (thoracic: ICC= 0.91 vs 0.51) and GN-MRA (root: ICC=0.84 vs 0.52; thoracic: ICC=0.93 vs 0.60). AGM-based 3D aortic size and growth maps were highly reproducible (median ICC >0.9 for diameters and >0.80 for GR).

CONCLUSION

Mapping aortic diameter and growth on MRA via 3D image registration is feasible, accurate and outperforms the current manual clinical standard. This technique could broaden the possibilities of clinical and research evaluation of patients with aortic thoracic diseases.

Mechanism Analysis of Vascular Calcification Based on Fluid Dynamics

Vascular calcification is the abnormal deposition of calcium phosphate complexes in blood vessels, which is regarded as the pathological basis of multiple cardiovascular diseases. The flowing blood exerts a frictional force called shear stress on the vascular wall. Blood vessels have different hydrodynamic properties due to discrepancies in geometric and mechanical properties. The disturbance of the blood flow in the bending area and the branch point of the arterial tree produces a shear stress lower than the physiological magnitude of the laminar shear stress, which can induce the occurrence of vascular calcification. Endothelial cells sense the fluid dynamics of blood and transmit electrical and chemical signals to the full-thickness of blood vessels. Through crosstalk with endothelial cells, smooth muscle cells trigger osteogenic transformation, involved in mediating vascular intima and media calcification. In addition, based on the detection of fluid dynamics parameters, emerging imaging technologies such as 4D Flow MRI and computational fluid dynamics have greatly improved the early diagnosis ability of cardiovascular diseases, showing extremely high clinical application prospects.

Automatic 4D Flow MRI Segmentation Using the Standardized Difference of Means Velocity

We present a method to automatically segment 4D flow magnetic resonance imaging (MRI) by identifying net flow effects using the standardized difference of means (SDM) velocity. The SDM velocity quantifies the ratio between the net flow and observed flow pulsatility in each voxel. Vessel segmentation is performed using an F-test, identifying voxels with significantly higher SDM velocity values than background voxels. We compare the SDM segmentation algorithm against pseudo-complex difference (PCD) intensity segmentation of 4D flow measurements in in vitro cerebral aneurysm models and 10 in vitro Circle of Willis (CoW) datasets. We also compared the SDM algorithm to convolutional neural network (CNN) segmentation in 5 thoracic vasculature datasets. The in vitro flow phantom geometry is known, while the ground truth geometries for the CoW and thoracic aortas are derived from high-resolution time-of-flight (TOF) magnetic resonance angiography and manual segmentation, respectively. The SDM algorithm demonstrates greater robustness than PCD and CNN approaches and can be applied to 4D flow data from other vascular territories. The SDM to PCD comparison demonstrated an approximate 48% increase in sensitivity in vitro and 70% increase in the CoW, respectively; the SDM and CNN sensitivities were similar. The vessel surface derived from the SDM method was 46% closer to the in vitro surfaces and 72% closer to the in vitro TOF surfaces than the PCD approach. The SDM and CNN approaches both accurately identify vessel surfaces. The SDM algorithm is a repeatable segmentation method, enabling reliable computation of hemodynamic metrics associated with cardiovascular disease.

Mechanisms of Aortic Dilation in Patients With Bicuspid Aortic Valve: JACC State-of-the-Art Review

Bicuspid aortic valve is the most common congenital heart disease and exposes patients to an increased risk of aortic dilation and dissection. Aortic dilation is a slow, silent process, leading to a greater risk of aortic dissection. The prevention of adverse events together with optimization of the frequency of the required lifelong imaging surveillance are important for both clinicians and patients and motivated extensive research to shed light on the physiopathologic processes involved in bicuspid aortic valve aortopathy. Two main research hypotheses have been consolidated in the last decade: one supports a genetic basis for the increased prevalence of dilation, in particular for the aortic root, and the second supports the damaging impact on the aortic wall of altered flow dynamics associated with these structurally abnormal valves, particularly significant in the ascending aorta. Current opinion tends to rule out mutually excluding causative mechanisms, recognizing both as important and potentially clinically relevant.

Impact of thoracic endovascular aortic repair following blunt traumatic thoracic aortic injury on blood pressure

BACKGROUND

Blunt traumatic thoracic aortic injuries (BTAIs) are associated with a high mortality rate. Thoracic endovascular aortic repair (TEVAR) is the most frequently used surgical strategy in patients with BTAI, as it offers good short- and middle-term results. Previous studies have reported an abnormally high prevalence of hypertension (HT) in these patients. This work aimed to describe the long-term prevalence of HT and provide a comprehensive evaluation of the biomechanical, clinical, and functional factors involved in HT development.

METHODS

Twenty-six patients treated with TEVAR following BTAI with no history of HT at the time of trauma were enrolled. They were matched with 37 healthy volunteers based on age, sex, and body surface area and underwent a comprehensive follow-up study, including cardiovascular magnetic resonance, 24-hour ambulatory blood pressure monitoring, and assessment of carotid-femoral pulse wave velocity (cfPWV, a measure of aortic stiffness) and flow-mediated vasodilation.

RESULTS

The mean patient age was 43.5 ± 12.9 years, and the majority were male (23 of 26; 88.5%). At a mean of 120.2 ± 69.7 months after intervention, 17 patients (65%) presented with HT, 14 (54%) had abnormal nighttime blood pressure dipping, and 6 (23%) high cfPWV. New-onset HT was related to a more proximal TEVAR landing zone and greater distal oversizing. Abnormal nighttime blood pressure was related to high cfPWV, which in turn was associated with TEVAR length and premature arterial aging.

CONCLUSIONS

HT frequently occurs otherwise healthy subjects undergoing TEVAR implantation after BTAI. TEVAR stiffness and length, the proximal landing zone, and distal oversizing are potentially modifiable surgical characteristics related to abnormal blood pressure.

Segmentation of 4D Flow MRI: Comparison between 3D Deep Learning and Velocity-Based Level Sets

A thoracic aortic aneurysm is an abnormal dilatation of the aorta that can progress and lead to rupture. The decision to conduct surgery is made by considering the maximum diameter, but it is now well known that this metric alone is not completely reliable. The advent of 4D flow magnetic resonance imaging has allowed for the calculation of new biomarkers for the study of aortic diseases, such as wall shear stress. However, the calculation of these biomarkers requires the precise segmentation of the aorta during all phases of the cardiac cycle. The objective of this work was to compare two different methods for automatically segmenting the thoracic aorta in the systolic phase using 4D flow MRI. The first method is based on a level set framework and uses the velocity field in addition to 3D phase contrast magnetic resonance imaging. The second method is a U-Net-like approach that is only applied to magnitude images from 4D flow MRI. The used dataset was composed of 36 exams from different patients, with ground truth data for the systolic phase of the cardiac cycle. The comparison was performed based on selected metrics, such as the Dice similarity coefficient (DSC) and Hausdorf distance (HD), for the whole aorta and also three aortic regions. Wall shear stress was also assessed and the maximum wall shear stress values were used for comparison. The U-Net-based approach provided statistically better results for the 3D segmentation of the aorta, with a DSC of 0.92 ± 0.02 vs. 0.86 ± 0.5 and an HD of 21.49 ± 24.8 mm vs. 35.79 ± 31.33 mm for the whole aorta. The absolute difference between the wall shear stress and ground truth slightly favored the level set method, but not significantly (0.754 ± 1.07 Pa vs. 0.737 ± 0.79 Pa). The results showed that the deep learning-based method should be considered for the segmentation of all time steps in order to evaluate biomarkers based on 4D flow MRI.

Magnetic Resonance Imaging and Computed Tomography for the Noninvasive Assessment of Arterial Aging: A Review by the VascAgeNet COST Action

Magnetic resonance imaging and computed tomography allow the characterization of arterial state and function with high confidence and thus play a key role in the understanding of arterial aging and its translation into the clinic. Decades of research into the development of innovative imaging sequences and image analysis techniques have led to the identification of a large number of potential biomarkers, some bringing improvement in basic science, others in clinical practice. Nonetheless, the complexity of some of these biomarkers and the image analysis techniques required for their computation hamper their widespread use. In this narrative review, current biomarkers related to aging of the aorta, their founding principles, the sequence, and postprocessing required, and their predictive values for cardiovascular events are summarized. For each biomarker a summary of reference values and reproducibility studies and limitations is provided. The present review, developed in the COST Action VascAgeNet, aims to guide clinicians and technical researchers in the critical understanding of the possibilities offered by these advanced imaging modalities for studying the state and function of the aorta, and their possible clinically relevant relationships with aging.

Advances in machine learning applications for cardiovascular 4D flow MRI

Four-dimensional flow magnetic resonance imaging (MRI) has evolved as a non-invasive imaging technique to visualize and quantify blood flow in the heart and vessels. Hemodynamic parameters derived from 4D flow MRI, such as net flow and peak velocities, but also kinetic energy, turbulent kinetic energy, viscous energy loss, and wall shear stress have shown to be of diagnostic relevance for cardiovascular diseases. 4D flow MRI, however, has several limitations. Its long acquisition times and its limited spatio-temporal resolutions lead to inaccuracies in velocity measurements in small and low-flow vessels and near the vessel wall. Additionally, 4D flow MRI requires long post-processing times, since inaccuracies due to the measurement process need to be corrected for and parameter quantification requires 2D and 3D contour drawing. Several machine learning (ML) techniques have been proposed to overcome these limitations. Existing scan acceleration methods have been extended using ML for image reconstruction and ML based super-resolution methods have been used to assimilate high-resolution computational fluid dynamic simulations and 4D flow MRI, which leads to more realistic velocity results. ML efforts have also focused on the automation of other post-processing steps, by learning phase corrections and anti-aliasing. To automate contour drawing and 3D segmentation, networks such as the U-Net have been widely applied. This review summarizes the latest ML advances in 4D flow MRI with a focus on technical aspects and applications. It is divided into the current status of fast and accurate 4D flow MRI data generation, ML based post-processing tools for phase correction and vessel delineation and the statistical evaluation of blood flow.