Fully-automated left ventricular mass and volume MRI analysis in the UK Biobank population cohort: evaluation of initial results

A novel deep learning method for large-scale analysis of bone marrow adiposity using UK Biobank Dixon MRI data

Background

Bone marrow adipose tissue (BMAT) represents > 10% fat mass in healthy humans and can be measured by magnetic resonance imaging (MRI) as the bone marrow fat fraction (BMFF). Human MRI studies have identified several diseases associated with BMFF but have been relatively small scale. Population-scale studies therefore have huge potential to reveal BMAT's true clinical relevance. The UK Biobank (UKBB) is undertaking MRI of 100,000 participants, providing the ideal opportunity for such advances.

Objective

To establish deep learning for high-throughput multi-site BMFF analysis from UKBB MRI data.

Materials and methods

We studied males and females aged 60-69. Bone marrow (BM) segmentation was automated using a new lightweight attention-based 3D U-Net convolutional neural network that improved segmentation of small structures from large volumetric data. Using manual segmentations from 61-64 subjects, the models were trained to segment four BM regions of interest: the spine (thoracic and lumbar vertebrae), femoral head, total hip and femoral diaphysis. Models were tested using a further 10-12 datasets per region and validated using datasets from 729 UKBB participants. BMFF was then quantified and pathophysiological characteristics assessed, including site- and sex-dependent differences and the relationships with age, BMI, bone mineral density, peripheral adiposity, and osteoporosis.

Results

Model accuracy matched or exceeded that for conventional U-Nets, yielding Dice scores of 91.2% (spine), 94.5% (femoral head), 91.2% (total hip) and 86.6% (femoral diaphysis). One case of severe scoliosis prevented segmentation of the spine, while one case of Non-Hodgkin Lymphoma prevented segmentation of the spine, femoral head and total hip because of T2 signal depletion; however, successful segmentation was not disrupted by any other pathophysiological variables. The resulting BMFF measurements confirmed expected relationships between BMFF and age, sex and bone density, and identified new site- and sex-specific characteristics.

Conclusions

We have established a new deep learning method for accurate segmentation of small structures from large volumetric data, allowing high-throughput multi-site BMFF measurement in the UKBB. Our findings reveal new pathophysiological insights, highlighting the potential of BMFF as a novel clinical biomarker. Applying our method across the full UKBB cohort will help to reveal the impact of BMAT on human health and disease.

A Multimodal Video-Based AI Biomarker for Aortic Stenosis Development and Progression

Importance

Aortic stenosis (AS) is a major public health challenge with a growing therapeutic landscape, but current biomarkers do not inform personalized screening and follow-up. A video-based artificial intelligence (AI) biomarker (Digital AS Severity index [DASSi]) can detect severe AS using single-view long-axis echocardiography without Doppler characterization.

Objective

To deploy DASSi to patients with no AS or with mild or moderate AS at baseline to identify AS development and progression.

Design, Setting, and Participants

This is a cohort study that examined 2 cohorts of patients without severe AS undergoing echocardiography in the Yale New Haven Health System (YNHHS; 2015-2021) and Cedars-Sinai Medical Center (CSMC; 2018-2019). A novel computational pipeline for the cross-modal translation of DASSi into cardiac magnetic resonance (CMR) imaging was further developed in the UK Biobank. Analyses were performed between August 2023 and February 2024.

Exposure

DASSi (range, 0-1) derived from AI applied to echocardiography and CMR videos.

Main Outcomes and Measures

Annualized change in peak aortic valve velocity (AV-Vmax) and late (>6 months) aortic valve replacement (AVR).

Results

A total of 12 599 participants were included in the echocardiographic study (YNHHS: n = 8798; median [IQR] age, 71 [60-80] years; 4250 [48.3%] women; median [IQR] follow-up, 4.1 [2.4-5.4] years; and CSMC: n = 3801; median [IQR] age, 67 [54-78] years; 1685 [44.3%] women; median [IQR] follow-up, 3.4 [2.8-3.9] years). Higher baseline DASSi was associated with faster progression in AV-Vmax (per 0.1 DASSi increment: YNHHS, 0.033 m/s per year [95% CI, 0.028-0.038] among 5483 participants; CSMC, 0.082 m/s per year [95% CI, 0.053-0.111] among 1292 participants), with values of 0.2 or greater associated with a 4- to 5-fold higher AVR risk than values less than 0.2 (YNHHS: 715 events; adjusted hazard ratio [HR], 4.97 [95% CI, 2.71-5.82]; CSMC: 56 events; adjusted HR, 4.04 [95% CI, 0.92-17.70]), independent of age, sex, race, ethnicity, ejection fraction, and AV-Vmax. This was reproduced across 45 474 participants (median [IQR] age, 65 [59-71] years; 23 559 [51.8%] women; median [IQR] follow-up, 2.5 [1.6-3.9] years) undergoing CMR imaging in the UK Biobank (for participants with DASSi ≥0.2 vs those with DASSi <.02, adjusted HR, 11.38 [95% CI, 2.56-50.57]). Saliency maps and phenome-wide association studies supported associations with cardiac structure and function and traditional cardiovascular risk factors.

Conclusions and Relevance

In this cohort study of patients without severe AS undergoing echocardiography or CMR imaging, a new AI-based video biomarker was independently associated with AS development and progression, enabling opportunistic risk stratification across cardiovascular imaging modalities as well as potential application on handheld devices.

Automatic left ventricle volume and mass quantification from 2D cine-MRI: Investigating papillary muscle influence

OBJECTIVE

Early detection of cardiovascular diseases is based on accurate quantification of the left ventricle (LV) function parameters. In this paper, we propose a fully automatic framework for LV volume and mass quantification from 2D-cine MR images already segmented using U-Net.

METHODS

The general framework consists of three main steps: Data preparation including automatic LV localization using a convolution neural network (CNN) and application of morphological operations to exclude papillary muscles from the LV cavity. The second step consists in automatically extracting the LV contours using U-Net architecture. Finally, by integrating temporal information which is manifested by a spatial motion of myocytes as a third dimension, we calculated LV volume, LV ejection fraction (LVEF) and left ventricle mass (LVM). Based on these parameters, we detected and quantified cardiac contraction abnormalities using Python software.

RESULTS

CNN was trained with 35 patients and tested on 15 patients from the ACDC database with an accuracy of 99,15 %. U-Net architecture was trained using ACDC database and evaluated using local dataset with a Dice similarity coefficient (DSC) of 99,78 % and a Hausdorff Distance (HD) of 4.468 mm (p < 0,001). Quantification results showed a strong correlation with physiological measures with a Pearson correlation coefficient (PCC) of 0,991 for LV volume, 0.962 for LVEF, 0.98 for stroke volume (SV) and 0.923 for LVM after pillars' elimination. Clinically, our method allows regional and accurate identification of pathological myocardial segments and can serve as a diagnostic aid tool of cardiac contraction abnormalities.

CONCLUSION

Experimental results prove the usefulness of the proposed method for LV volume and function quantification and verify its potential clinical applicability.

Correcting bias in cardiac geometries derived from multimodal images using spatiotemporal mapping

Cardiovascular imaging studies provide a multitude of structural and functional data to better understand disease mechanisms. While pooling data across studies enables more powerful and broader applications, performing quantitative comparisons across datasets with varying acquisition or analysis methods is problematic due to inherent measurement biases specific to each protocol. We show how dynamic time warping and partial least squares regression can be applied to effectively map between left ventricular geometries derived from different imaging modalities and analysis protocols to account for such differences. To demonstrate this method, paired real-time 3D echocardiography (3DE) and cardiac magnetic resonance (CMR) sequences from 138 subjects were used to construct a mapping function between the two modalities to correct for biases in left ventricular clinical cardiac indices, as well as regional shape. Leave-one-out cross-validation revealed a significant reduction in mean bias, narrower limits of agreement, and higher intraclass correlation coefficients for all functional indices between CMR and 3DE geometries after spatiotemporal mapping. Meanwhile, average root mean squared errors between surface coordinates of 3DE and CMR geometries across the cardiac cycle decreased from 7 ± 1 to 4 ± 1 mm for the total study population. Our generalised method for mapping between time-varying cardiac geometries obtained using different acquisition and analysis protocols enables the pooling of data between modalities and the potential for smaller studies to leverage large population databases for quantitative comparisons.

Artificial intelligence in cardiovascular imaging: enhancing image analysis and risk stratification

In this review, we summarize state-of-the-art artificial intelligence applications for non-invasive cardiovascular imaging modalities including CT, MRI, echocardiography, and nuclear myocardial perfusion imaging.

Clinical and genetic associations of deep learning-derived cardiac magnetic resonance-based left ventricular mass

Left ventricular mass is a risk marker for cardiovascular events, and may indicate an underlying cardiomyopathy. Cardiac magnetic resonance is the gold-standard for left ventricular mass estimation, but is challenging to obtain at scale. Here, we use deep learning to enable genome-wide association study of cardiac magnetic resonance-derived left ventricular mass indexed to body surface area within 43,230 UK Biobank participants. We identify 12 genome-wide associations (1 known at TTN and 11 novel for left ventricular mass), implicating genes previously associated with cardiac contractility and cardiomyopathy. Cardiac magnetic resonance-derived indexed left ventricular mass is associated with incident dilated and hypertrophic cardiomyopathies, and implantable cardioverter-defibrillator implant. An indexed left ventricular mass polygenic risk score ≥90th percentile is also associated with incident implantable cardioverter-defibrillator implant in separate UK Biobank (hazard ratio 1.22, 95% CI 1.05-1.44) and Mass General Brigham (hazard ratio 1.75, 95% CI 1.12-2.74) samples. Here, we perform a genome-wide association study of cardiac magnetic resonance-derived indexed left ventricular mass to identify 11 novel variants and demonstrate that cardiac magnetic resonance-derived and genetically predicted indexed left ventricular mass are associated with incident cardiomyopathy.

Impact of Measurement Imprecision on Genetic Association Studies of Cardiac Function

Background

Recent studies have leveraged quantitative traits from imaging to amplify the power of genome-wide association studies (GWAS) to gain further insights into the biology of diseases and traits. However, measurement imprecision is intrinsic to phenotyping and can impact downstream genetic analyses.

Methods

Left ventricular ejection fraction (LVEF), an important but imprecise quantitative imaging measurement, was examined to assess the impact of precision of phenotype measurement on genetic studies. Multiple approaches to obtain LVEF, as well as simulated measurement noise, were evaluated with their impact on downstream genetic analyses.

Results

Even within the same population, small changes in the measurement of LVEF drastically impacted downstream genetic analyses. Introducing measurement noise as little as 7.9% can eliminate all significant genetic associations in an GWAS with almost forty thousand individuals. An increase of 1% in mean absolute error (MAE) in LVEF had an equivalent impact on GWAS power as a decrease of 10% in the cohort sample size, suggesting optimizing phenotyping precision is a cost-effective way to improve power of genetic studies.

Conclusions

Improving the precision of phenotyping is important for maximizing the yield of genome-wide association studies.

Frequency, Penetrance, and Variable Expressivity of Dilated Cardiomyopathy-Associated Putative Pathogenic Gene Variants in UK Biobank Participants

BACKGROUND

There is a paucity of data regarding the phenotype of dilated cardiomyopathy (DCM) gene variants in the general population. We aimed to determine the frequency and penetrance of DCM-associated putative pathogenic gene variants in a general adult population, with a focus on the expression of clinical and subclinical phenotype, including structural, functional, and arrhythmic disease features.

METHODS

UK Biobank participants who had undergone whole exome sequencing, ECG, and cardiovascular magnetic resonance imaging were selected for study. Three variant-calling strategies (1 primary and 2 secondary) were used to identify participants with putative pathogenic variants in 44 DCM genes. The observed phenotype was graded DCM (clinical or cardiovascular magnetic resonance diagnosis); early DCM features, including arrhythmia or conduction disease, isolated ventricular dilation, and hypokinetic nondilated cardiomyopathy; or phenotype-negative.

RESULTS

Among 18 665 individuals included in the study, 1463 (7.8%) possessed ≥1 putative pathogenic variant in 44 DCM genes by the main variant calling strategy. A clinical diagnosis of DCM was present in 0.34% and early DCM features in 5.7% of individuals with putative pathogenic variants. ECG and cardiovascular magnetic resonance analysis revealed evidence of subclinical DCM in an additional 1.6% and early DCM features in an additional 15.9% of individuals with putative pathogenic variants. Arrhythmias or conduction disease (15.2%) were the most common early DCM features, followed by hypokinetic nondilated cardiomyopathy (4%). The combined clinical/subclinical penetrance was ≤30% with all 3 variant filtering strategies. Clinical DCM was slightly more prevalent among participants with putative pathogenic variants in definitive/strong evidence genes as compared with those with variants in moderate/limited evidence genes.

CONCLUSIONS

In the UK Biobank, ≈1 of 6 of adults with putative pathogenic variants in DCM genes exhibited early DCM features potentially associated with DCM genotype, most commonly manifesting with arrhythmias in the absence of substantial ventricular dilation or dysfunction.

Training and clinical testing of artificial intelligence derived right atrial cardiovascular magnetic resonance measurements

BACKGROUND

Right atrial (RA) area predicts mortality in patients with pulmonary hypertension, and is recommended by the European Society of Cardiology/European Respiratory Society pulmonary hypertension guidelines. The advent of deep learning may allow more reliable measurement of RA areas to improve clinical assessments. The aim of this study was to automate cardiovascular magnetic resonance (CMR) RA area measurements and evaluate the clinical utility by assessing repeatability, correlation with invasive haemodynamics and prognostic value.

METHODS

A deep learning RA area CMR contouring model was trained in a multicentre cohort of 365 patients with pulmonary hypertension, left ventricular pathology and healthy subjects. Inter-study repeatability (intraclass correlation coefficient (ICC)) and agreement of contours (DICE similarity coefficient (DSC)) were assessed in a prospective cohort (n = 36). Clinical testing and mortality prediction was performed in n = 400 patients that were not used in the training nor prospective cohort, and the correlation of automatic and manual RA measurements with invasive haemodynamics assessed in n = 212/400. Radiologist quality control (QC) was performed in the ASPIRE registry, n = 3795 patients. The primary QC observer evaluated all the segmentations and recorded them as satisfactory, suboptimal or failure. A second QC observer analysed a random subcohort to assess QC agreement (n = 1018).

RESULTS

All deep learning RA measurements showed higher interstudy repeatability (ICC 0.91 to 0.95) compared to manual RA measurements (1st observer ICC 0.82 to 0.88, 2nd observer ICC 0.88 to 0.91). DSC showed high agreement comparing automatic artificial intelligence and manual CMR readers. Maximal RA area mean and standard deviation (SD) DSC metric for observer 1 vs observer 2, automatic measurements vs observer 1 and automatic measurements vs observer 2 is 92.4 ± 3.5 cm2, 91.2 ± 4.5 cm2 and 93.2 ± 3.2 cm2, respectively. Minimal RA area mean and SD DSC metric for observer 1 vs observer 2, automatic measurements vs observer 1 and automatic measurements vs observer 2 was 89.8 ± 3.9 cm2, 87.0 ± 5.8 cm2 and 91.8 ± 4.8 cm2. Automatic RA area measurements all showed moderate correlation with invasive parameters (r = 0.45 to 0.66), manual (r = 0.36 to 0.57). Maximal RA area could accurately predict elevated mean RA pressure low and high-risk thresholds (area under the receiver operating characteristic curve artificial intelligence = 0.82/0.87 vs manual = 0.78/0.83), and predicted mortality similar to manual measurements, both p < 0.01. In the QC evaluation, artificial intelligence segmentations were suboptimal at 108/3795 and a low failure rate of 16/3795. In a subcohort (n = 1018), agreement by two QC observers was excellent, kappa 0.84.

CONCLUSION

Automatic artificial intelligence CMR derived RA size and function are accurate, have excellent repeatability, moderate associations with invasive haemodynamics and predict mortality.

AI Based CMR Assessment of Biventricular Function: Clinical Significance of Intervendor Variability and Measurement Errors

OBJECTIVES

The aim of this study was to determine whether left ventricular ejection fraction (LVEF) and right ventricular ejection fraction (RVEF) and left ventricular mass (LVM) measurements made using 3 fully automated deep learning (DL) algorithms are accurate and interchangeable and can be used to classify ventricular function and risk-stratify patients as accurately as an expert.

BACKGROUND

Artificial intelligence is increasingly used to assess cardiac function and LVM from cardiac magnetic resonance images.

METHODS

Two hundred patients were identified from a registry of individuals who underwent vasodilator stress cardiac magnetic resonance. LVEF, LVM, and RVEF were determined using 3 fully automated commercial DL algorithms and by a clinical expert (CLIN) using conventional methodology. Additionally, LVEF values were classified according to clinically important ranges: <35%, 35% to 50%, and ≥50%. Both ejection fraction values and classifications made by the DL ejection fraction approaches were compared against CLIN ejection fraction reference. Receiver-operating characteristic curve analysis was performed to evaluate the ability of CLIN and each of the DL classifications to predict major adverse cardiovascular events.

RESULTS

Excellent correlations were seen for each DL-LVEF compared with CLIN-LVEF (r = 0.83-0.93). Good correlations were present between DL-LVM and CLIN-LVM (r = 0.75-0.85). Modest correlations were observed between DL-RVEF and CLIN-RVEF (r = 0.59-0.68). A >10% error between CLIN and DL ejection fraction was present in 5% to 18% of cases for the left ventricle and 23% to 43% for the right ventricle. LVEF classification agreed with CLIN-LVEF classification in 86%, 80%, and 85% cases for the 3 DL-LVEF approaches. There were no differences among the 4 approaches in associations with major adverse cardiovascular events for LVEF, LVM, and RVEF.

CONCLUSIONS

This study revealed good agreement between automated and expert-derived LVEF and similarly strong associations with outcomes, compared with an expert. However, the ability of these automated measurements to accurately classify left ventricular function for treatment decision remains limited. DL-LVM showed good agreement with CLIN-LVM. DL-RVEF approaches need further refinements.

Balancing Speed and Accuracy in Cardiac Magnetic Resonance Function Post-Processing: Comparing 2 Levels of Automation in 3 Vendors to Manual Assessment

Automating cardiac function assessment on cardiac magnetic resonance short-axis cines is faster and more reproducible than manual contour-tracing; however, accurately tracing basal contours remains challenging. Three automated post-processing software packages (Level 1) were compared to manual assessment. Subsequently, automated basal tracings were manually adjusted using a standardized protocol combined with software package-specific relative-to-manual standard error correction (Level 2). All post-processing was performed in 65 healthy subjects. Manual contour-tracing was performed separately from Level 1 and 2 automated analysis. Automated measurements were considered accurate when the difference was equal or less than the maximum manual inter-observer disagreement percentage. Level 1 (2.1 ± 1.0 min) and Level 2 automated (5.2 ± 1.3 min) were faster and more reproducible than manual (21.1 ± 2.9 min) post-processing, the maximum inter-observer disagreement was 6%. Compared to manual, Level 1 automation had wide limits of agreement. The most reliable software package obtained more accurate measurements in Level 2 compared to Level 1 automation: left ventricular end-diastolic volume, 98% and 53%; ejection fraction, 98% and 60%; mass, 70% and 3%; right ventricular end-diastolic volume, 98% and 28%; ejection fraction, 80% and 40%, respectively. Level 1 automated cardiac function post-processing is fast and highly reproducible with varying accuracy. Level 2 automation balances speed and accuracy.

Effect of age and sex on fully automated deep learning assessment of left ventricular function, volumes, and contours in cardiac magnetic resonance imaging

Deep learning algorithms for left ventricle (LV) segmentation are prone to bias towards the training dataset. This study assesses sex- and age-dependent performance differences when using deep learning for automatic LV segmentation. Retrospective analysis of 100 healthy subjects undergoing cardiac MRI from 2012 to 2018, with 10 men and women in the following age groups: 18-30, 31-40, 41-50, 51-60, and 61-80 years old. Subjects underwent 1.5 T, 2D CINE SSFP MRI. 35 pathologic cases from local clinical exams and the SCMR 2015 consensus contours dataset were also analyzed. A fully convolutional network (FCN) similar to U-Net trained on the U.K. Biobank was used to automatically segment LV endocardial and epicardial contours. FCN and manual segmentation were compared using Dice metrics and measurements of end-diastolic volume (EDV), end-systolic volume (ESV), mass (LVM), and ejection fraction (LVEF). Paired t-tests and linear regressions were used to analyze measurement differences with respect to sex and age. Dice metrics (median ± IQR) for n = 135 cases were 0.94 ± 0.04/0.87 ± 0.10 (ED endocardium/ES endocardium). Measurement biases (mean ± SD) among the healthy cohort were - 0.3 ± 10.1 mL for EDV, - 6.7 ± 9.6 mL for ESV, 4.6 ± 6.4% for LVEF, and - 2.2 ± 11.0 g for LVM; biases were independent of sex and age. Biases among the 35 pathologic cases were 0.1 ± 19 mL for EDV, - 4.8 ± 19 mL for ESV, 2.0 ± 7.6% for LVEF, and 1.0 ± 20 g for LVM. In conclusion, automatic segmentation by the Biobank-trained FCN was independent of age and sex. Improvements in end-systolic basal slice detection are needed to decrease bias and improve precision in ESV and LVEF.

The Promise of AI in Detection, Diagnosis, and Epidemiology for Combating COVID-19: Beyond the Hype

COVID-19 has created enormous suffering, affecting lives, and causing deaths. The ease with which this type of coronavirus can spread has exposed weaknesses of many healthcare systems around the world. Since its emergence, many governments, research communities, commercial enterprises, and other institutions and stakeholders around the world have been fighting in various ways to curb the spread of the disease. Science and technology have helped in the implementation of policies of many governments that are directed toward mitigating the impacts of the pandemic and in diagnosing and providing care for the disease. Recent technological tools, artificial intelligence (AI) tools in particular, have also been explored to track the spread of the coronavirus, identify patients with high mortality risk and diagnose patients for the disease. In this paper, areas where AI techniques are being used in the detection, diagnosis and epidemiological predictions, forecasting and social control for combating COVID-19 are discussed, highlighting areas of successful applications and underscoring issues that need to be addressed to achieve significant progress in battling COVID-19 and future pandemics. Several AI systems have been developed for diagnosing COVID-19 using medical imaging modalities such as chest CT and X-ray images. These AI systems mainly differ in their choices of the algorithms for image segmentation, classification and disease diagnosis. Other AI-based systems have focused on predicting mortality rate, long-term patient hospitalization and patient outcomes for COVID-19. AI has huge potential in the battle against the COVID-19 pandemic but successful practical deployments of these AI-based tools have so far been limited due to challenges such as limited data accessibility, the need for external evaluation of AI models, the lack of awareness of AI experts of the regulatory landscape governing the deployment of AI tools in healthcare, the need for clinicians and other experts to work with AI experts in a multidisciplinary context and the need to address public concerns over data collection, privacy, and protection. Having a dedicated team with expertise in medical data collection, privacy, access and sharing, using federated learning whereby AI scientists hand over training algorithms to the healthcare institutions to train models locally, and taking full advantage of biomedical data stored in biobanks can alleviate some of problems posed by these challenges. Addressing these challenges will ultimately accelerate the translation of AI research into practical and useful solutions for combating pandemics.

Deep learning to estimate cardiac magnetic resonance-derived left ventricular mass

Background

Cardiac magnetic resonance (CMR) is the gold standard for left ventricular hypertrophy (LVH) diagnosis. CMR-derived LV mass can be estimated using proprietary algorithms (eg, InlineVF), but their accuracy and availability may be limited.

Objective

To develop an open-source deep learning model to estimate CMR-derived LV mass.

Methods

Within participants of the UK Biobank prospective cohort undergoing CMR, we trained 2 convolutional neural networks to estimate LV mass. The first (ML4Hreg) performed regression informed by manually labeled LV mass (available in 5065 individuals), while the second (ML4Hseg) performed LV segmentation informed by InlineVF (version D13A) contours. We compared ML4Hreg, ML4Hseg, and InlineVF against manually labeled LV mass within an independent holdout set using Pearson correlation and mean absolute error (MAE). We assessed associations between CMR-derived LVH and prevalent cardiovascular disease using logistic regression adjusted for age and sex.

Results

We generated CMR-derived LV mass estimates within 38,574 individuals. Among 891 individuals in the holdout set, ML4Hseg reproduced manually labeled LV mass more accurately (r = 0.864, 95% confidence interval [CI] 0.847-0.880; MAE 10.41 g, 95% CI 9.82-10.99) than ML4Hreg (r = 0.843, 95% CI 0.823-0.861; MAE 10.51, 95% CI 9.86-11.15, P = .01) and InlineVF (r = 0.795, 95% CI 0.770-0.818; MAE 14.30, 95% CI 13.46-11.01, P < .01). LVH defined using ML4Hseg demonstrated the strongest associations with hypertension (odds ratio 2.76, 95% CI 2.51-3.04), atrial fibrillation (1.75, 95% CI 1.37-2.20), and heart failure (4.67, 95% CI 3.28-6.49).

Conclusions

ML4Hseg is an open-source deep learning model providing automated quantification of CMR-derived LV mass. Deep learning models characterizing cardiac structure may facilitate broad cardiovascular discovery.

Cardiovascular magnetic resonance imaging in the UK Biobank: a major international health research resource

The UK Biobank (UKB) is a health research resource of major international importance, incorporating comprehensive characterization of >500 000 men and women recruited between 2006 and 2010 from across the UK. There is prospective tracking of health outcomes for all participants through linkages with national cohorts (death registers, cancer registers, electronic hospital records, and primary care records). The dataset has been enhanced with the UKB imaging study, which aims to scan a subset of 100 000 participants. The imaging protocol includes magnetic resonance imaging of the brain, heart, and abdomen, carotid ultrasound, and whole-body dual X-ray absorptiometry. Since its launch in 2015, over 48 000 participants have completed the imaging study with scheduled completion in 2023. Repeat imaging of 10 000 participants has been approved and commenced in 2019. The cardiovascular magnetic resonance (CMR) scan provides detailed assessment of cardiac structure and function comprising bright blood anatomic assessment (sagittal, coronal, and axial), left and right ventricular cine images (long and short axes), myocardial tagging, native T1 mapping, aortic flow, and imaging of the thoracic aorta. The UKB is an open access resource available to health researchers across all scientific disciplines from both academia and industry with no preferential access or exclusivity. In this paper, we consider how we may best utilize the UKB CMR data to advance cardiovascular research and review notable achievements to date.

Retraining Convolutional Neural Networks for Specialized Cardiovascular Imaging Tasks: Lessons from Tetralogy of Fallot

Ventricular contouring of cardiac magnetic resonance imaging is the gold standard for volumetric analysis for repaired tetralogy of Fallot (rTOF), but can be time-consuming and subject to variability. A convolutional neural network (CNN) ventricular contouring algorithm was developed to generate contours for mostly structural normal hearts. We aimed to improve this algorithm for use in rTOF and propose a more comprehensive method of evaluating algorithm performance. We evaluated the performance of a ventricular contouring CNN, that was trained on mostly structurally normal hearts, on rTOF patients. We then created an updated CNN by adding rTOF training cases and evaluated the new algorithm's performance generating contours for both the left and right ventricles (LV and RV) on new testing data. Algorithm performance was evaluated with spatial metrics (Dice Similarity Coefficient (DSC), Hausdorff distance, and average Hausdorff distance) and volumetric comparisons (e.g., differences in RV volumes). The original Mostly Structurally Normal (MSN) algorithm was better at contouring the LV than the RV in patients with rTOF. After retraining the algorithm, the new MSN + rTOF algorithm showed improvements for LV epicardial and RV endocardial contours on testing data to which it was naïve (N = 30; e.g., DSC 0.883 vs. 0.905 for LV epicardium at end diastole, p < 0.0001) and improvements in RV end-diastolic volumetrics (median %error 8.1 vs 11.4, p = 0.0022). Even with a small number of cases, CNN-based contouring for rTOF can be improved. This work should be extended to other forms of congenital heart disease with more extreme structural abnormalities. Aspects of this work have already been implemented in clinical practice, representing rapid clinical translation. The combined use of both spatial and volumetric comparisons yielded insights into algorithm errors.

Pathological Cluster Identification by Unsupervised Analysis in 3,822 UK Biobank Cardiac MRIs

We perform unsupervised analysis of image-derived shape and motion features extracted from 3,822 cardiac Magnetic resonance imaging (MRIs) of the UK Biobank. First, with a feature extraction method previously published based on deep learning models, we extract from each case 9 feature values characterizing both the cardiac shape and motion. Second, a feature selection is performed to remove highly correlated feature pairs. Third, clustering is carried out using a Gaussian mixture model on the selected features. After analysis, we identify 2 small clusters that probably correspond to 2 pathological categories. Further confirmation using a trained classification model and dimensionality reduction tools is carried out to support this finding. Moreover, we examine the differences between the other large clusters and compare our measures with the ground truth.

L-CO-Net: Learned Condensation-Optimization Network for Segmentation and Clinical Parameter Estimation from Cardiac Cine MRI

In this work, we implement a fully convolutional segmenter featuring both a learned group structure and a regularized weight-pruner to reduce the high computational cost in volumetric image segmentation. We validated our framework on the ACDC dataset featuring one healthy and four pathology patient groups imaged throughout the cardiac cycle. Our technique achieved Dice scores of 96.8% (LV blood-pool), 93.3% (RV blood-pool), and 90.0% (LV Myocardium) with five-fold cross-validation and yielded similar clinical parameters as those estimated from the ground-truth segmentation data. Based on these results, this technique has the potential to become an efficient and competitive cardiac image segmentation tool that may be used for cardiac computer-aided diagnosis, planning, and guidance applications.

PTPRG is an ischemia risk locus essential for HCO3--dependent regulation of endothelial function and tissue perfusion

Acid-base conditions modify artery tone and tissue perfusion but the involved vascular-sensing mechanisms and disease consequences remain unclear. We experimentally investigated transgenic mice and performed genetic studies in a UK-based human cohort. We show that endothelial cells express the putative HCO3--sensor receptor-type tyrosine-protein phosphatase RPTPγ, which enhances endothelial intracellular Ca2+-responses in resistance arteries and facilitates endothelium-dependent vasorelaxation only when CO2/HCO3- is present. Consistent with waning RPTPγ-dependent vasorelaxation at low [HCO3-], RPTPγ limits increases in cerebral perfusion during neuronal activity and augments decreases in cerebral perfusion during hyperventilation. RPTPγ does not influence resting blood pressure but amplifies hyperventilation-induced blood pressure elevations. Loss-of-function variants in PTPRG, encoding RPTPγ, are associated with increased risk of cerebral infarction, heart attack, and reduced cardiac ejection fraction. We conclude that PTPRG is an ischemia susceptibility locus; and RPTPγ-dependent sensing of HCO3- adjusts endothelium-mediated vasorelaxation, microvascular perfusion, and blood pressure during acid-base disturbances and altered tissue metabolism.

Fully automated quantification of left ventricular volumes and function in cardiac MRI: clinical evaluation of a deep learning-based algorithm

To investigate the performance of a deep learning-based algorithm for fully automated quantification of left ventricular (LV) volumes and function in cardiac MRI. We retrospectively analysed MR examinations of 50 patients (74% men, median age 57 years). The most common indications were known or suspected ischemic heart disease, cardiomyopathies or myocarditis. Fully automated analysis of LV volumes and function was performed using a deep learning-based algorithm. The analysis was subsequently corrected by a senior cardiovascular radiologist. Manual volumetric analysis was performed by two radiology trainees. Volumetric results were compared using Bland–Altman statistics and intra-class correlation coefficient. The frequency of clinically relevant differences was analysed using re-classification rates. The fully automated volumetric analysis was completed in a median of 8 s. With expert review and corrections, the analysis required a median of 110 s. Median time required for manual analysis was 3.5 min for a cardiovascular imaging fellow and 9 min for a radiology resident (p < 0.0001 for all comparisons). The correlation between fully automated results and expert-corrected results was very strong with intra-class correlation coefficients of 0.998 for end-diastolic volume, 0.997 for end-systolic volume, 0.899 for stroke volume, 0.972 for ejection fraction and 0.991 for myocardial mass (all p < 0.001). Clinically meaningful differences between fully automated and expert corrected results occurred in 18% of cases, comparable to the rate between the two manual readers (20%). Deep learning-based fully automated analysis of LV volumes and function is feasible, time-efficient and highly accurate. Clinically relevant corrections are required in a minority of cases.

Artificial intelligence: improving the efficiency of cardiovascular imaging

INTRODUCTION

Artificial intelligence (AI) describes the use of computational techniques to mimic human intelligence. In healthcare, this typically involves large medical datasets being used to predict a diagnosis, identify new disease genotypes or phenotypes, or guide treatment strategies. Noninvasive imaging remains a cornerstone for the diagnosis, risk stratification, and management of patients with cardiovascular disease. AI can facilitate every stage of the imaging process, from acquisition and reconstruction, to segmentation, measurement, interpretation, and subsequent clinical pathways.

AREAS COVERED

In this paper, we review state-of-the-art AI techniques and their current applications in cardiac imaging, and discuss the future role of AI as a precision medicine tool.

EXPERT OPINION

Cardiovascular medicine is primed for scalable AI applications which can interpret vast amounts of clinical and imaging data in greater depth than ever before. AI-augmented medical systems have the potential to improve workflow and provide reproducible and objective quantitative results which can inform clinical decisions. In the foreseeable future, AI may work in the background of cardiac image analysis software and routine clinical reporting, automatically collecting data and enabling real-time diagnosis and risk stratification.

The UK Biobank imaging enhancement of 100,000 participants: rationale, data collection, management and future directions

UK Biobank is a population-based cohort of half a million participants aged 40-69 years recruited between 2006 and 2010. In 2014, UK Biobank started the world's largest multi-modal imaging study, with the aim of re-inviting 100,000 participants to undergo brain, cardiac and abdominal magnetic resonance imaging, dual-energy X-ray absorptiometry and carotid ultrasound. The combination of large-scale multi-modal imaging with extensive phenotypic and genetic data offers an unprecedented resource for scientists to conduct health-related research. This article provides an in-depth overview of the imaging enhancement, including the data collected, how it is managed and processed, and future directions.

A deep learning-based approach for automatic segmentation and quantification of the left ventricle from cardiac cine MR images

Cardiac MRI has been widely used for noninvasive assessment of cardiac anatomy and function as well as heart diagnosis. The estimation of physiological heart parameters for heart diagnosis essentially require accurate segmentation of the Left ventricle (LV) from cardiac MRI. Therefore, we propose a novel deep learning approach for the automated segmentation and quantification of the LV from cardiac cine MR images. We aim to achieve lower errors for the estimated heart parameters compared to the previous studies by proposing a novel deep learning segmentation method. Our framework starts by an accurate localization of the LV blood pool center-point using a fully convolutional neural network (FCN) architecture called FCN1. Then, a region of interest (ROI) that contains the LV is extracted from all heart sections. The extracted ROIs are used for the segmentation of LV cavity and myocardium via a novel FCN architecture called FCN2. The FCN2 network has several bottleneck layers and uses less memory footprint than conventional architectures such as U-net. Furthermore, a new loss function called radial loss that minimizes the distance between the predicted and true contours of the LV is introduced into our model. Following myocardial segmentation, functional and mass parameters of the LV are estimated. Automated Cardiac Diagnosis Challenge (ACDC-2017) dataset was used to validate our framework, which gave better segmentation, accurate estimation of cardiac parameters, and produced less error compared to other methods applied on the same dataset. Furthermore, we showed that our segmentation approach generalizes well across different datasets by testing its performance on a locally acquired dataset. To sum up, we propose a deep learning approach that can be translated into a clinical tool for heart diagnosis.

Artificial Intelligence for Cardiac Imaging-Genetics Research

Cardiovascular conditions remain the leading cause of mortality and morbidity worldwide, with genotype being a significant influence on disease risk. Cardiac imaging-genetics aims to identify and characterize the genetic variants that influence functional, physiological, and anatomical phenotypes derived from cardiovascular imaging. High-throughput DNA sequencing and genotyping have greatly accelerated genetic discovery, making variant interpretation one of the key challenges in contemporary clinical genetics. Heterogeneous, low-fidelity phenotyping and difficulties integrating and then analyzing large-scale genetic, imaging and clinical datasets using traditional statistical approaches have impeded process. Artificial intelligence (AI) methods, such as deep learning, are particularly suited to tackle the challenges of scalability and high dimensionality of data and show promise in the field of cardiac imaging-genetics. Here we review the current state of AI as applied to imaging-genetics research and discuss outstanding methodological challenges, as the field moves from pilot studies to mainstream applications, from one dimensional global descriptors to high-resolution models of whole-organ shape and function, from univariate to multivariate analysis and from candidate gene to genome-wide approaches. Finally, we consider the future directions and prospects of AI imaging-genetics for ultimately helping understand the genetic and environmental underpinnings of cardiovascular health and disease.

Comparison of quality control methods for automated diffusion tensor imaging analysis pipelines

The processing of brain diffusion tensor imaging (DTI) data for large cohort studies requires fully automatic pipelines to perform quality control (QC) and artifact/outlier removal procedures on the raw DTI data prior to calculation of diffusion parameters. In this study, three automatic DTI processing pipelines, each complying with the general ENIGMA framework, were designed by uniquely combining multiple image processing software tools. Different QC procedures based on the RESTORE algorithm, the DTIPrep protocol, and a combination of both methods were compared using simulated ground truth and artifact containing DTI datasets modeling eddy current induced distortions, various levels of motion artifacts, and thermal noise. Variability was also examined in 20 DTI datasets acquired in subjects with vascular cognitive impairment (VCI) from the multi-site Ontario Neurodegenerative Disease Research Initiative (ONDRI). The mean fractional anisotropy (FA), mean diffusivity (MD), axial diffusivity (AD), and radial diffusivity (RD) were calculated in global brain grey matter (GM) and white matter (WM) regions. For the simulated DTI datasets, the measure used to evaluate the performance of the pipelines was the normalized difference between the mean DTI metrics measured in GM and WM regions and the corresponding ground truth DTI value. The performance of the proposed pipelines was very similar, particularly in FA measurements. However, the pipeline based on the RESTORE algorithm was the most accurate when analyzing the artifact containing DTI datasets. The pipeline that combined the DTIPrep protocol and the RESTORE algorithm produced the lowest standard deviation in FA measurements in normal appearing WM across subjects. We concluded that this pipeline was the most robust and is preferred for automated analysis of multisite brain DTI data.

Standardized image post-processing of cardiovascular magnetic resonance T1-mapping reduces variability and improves accuracy and consistency in myocardial tissue characterization

BACKGROUND

Myocardial T1-mapping is increasingly used in multicentre studies and trials. Inconsistent image analysis introduces variability, hinders differentiation of diseases, and results in larger sample sizes. We present a systematic approach to standardize T1-map analysis by human operators to improve accuracy and consistency.

METHODS

We developed a multi-step training program for T1-map post-processing. The training dataset contained 42 left ventricular (LV) short-axis T1-maps (normal and diseases; 1.5 and 3 Tesla). Contours drawn by two experienced human operators served as reference for myocardial T1 and wall thickness (WT). Trainees (n = 26) underwent training and were evaluated by: (a) qualitative review of contours; (b) quantitative comparison with reference T1 and WT.

RESULTS

The mean absolute difference between reference operators was 8.4 ± 6.3 ms (T1) and 1.2 ± 0.7 pixels (WT). Trainees' mean discrepancy from reference in T1 improved significantly post-training (from 8.1 ± 2.4 to 6.7 ± 1.4 ms; p < 0.001), with a 43% reduction in standard deviation (SD) (p = 0.035). WT also improved significantly post-training (from 0.9 ± 0.4 to 0.7 ± 0.2 pixels, p = 0.036), with 47% reduction in SD (p = 0.04). These experimentally-derived thresholds served to guide the training process: T1 (±8 ms) and WT (±1 pixel) from reference.

CONCLUSION

A standardized approach to CMR T1-map image post-processing leads to significant improvements in the accuracy and consistency of LV myocardial T1 values and wall thickness. Improving consistency between operators can translate into 33-72% reduction in clinical trial sample-sizes. This work may: (a) serve as a basis for re-certification for core-lab operators; (b) translate to sample-size reductions for clinical studies; (c) produce better-quality training datasets for machine learning.

Artificial intelligence in cardiovascular imaging: state of the art and implications for the imaging cardiologist

Healthcare, conceivably more than any other area of human endeavour, has the greatest potential to be affected by artificial intelligence (AI). This potential has been shown by several reports that demonstrate equal or superhuman performance in medical tasks that aim to improve efficiency, diagnosis and prognosis. This review focuses on the state of the art of AI applications in cardiovascular imaging. It provides an overview of the current applications and studies performed, including the potential value, implications, limitations and future directions of AI in cardiovascular imaging.It is envisioned that AI will dramatically change the way doctors practise medicine. In the short term, it will assist physicians with easy tasks, such as automating measurements, making predictions based on big data, and putting clinical findings into an evidence-based context. In the long term, AI will not only assist doctors, it has the potential to significantly improve access to health and well-being data for patients and their caretakers. This empowers patients. From a physician's perspective, reliable AI assistance will be available to support clinical decision-making. Although cardiovascular studies implementing AI are increasing in number, the applications have only just started to penetrate contemporary clinical care.

Fully Automated, Quality-Controlled Cardiac Analysis From CMR: Validation and Large-Scale Application to Characterize Cardiac Function

Objectives

This study sought to develop a fully automated framework for cardiac function analysis from cardiac magnetic resonance (CMR), including comprehensive quality control (QC) algorithms to detect erroneous output.

Background

Analysis of cine CMR imaging using deep learning (DL) algorithms could automate ventricular function assessment. However, variable image quality, variability in phenotypes of disease, and unavoidable weaknesses in training of DL algorithms currently prevent their use in clinical practice.

Methods

The framework consists of a pre-analysis DL image QC, followed by a DL algorithm for biventricular segmentation in long-axis and short-axis views, myocardial feature-tracking (FT), and a post-analysis QC to detect erroneous results. The study validated the framework in healthy subjects and cardiac patients by comparison against manual analysis (n = 100) and evaluation of the QC steps’ ability to detect erroneous results (n = 700). Next, this method was used to obtain reference values for cardiac function metrics from the UK Biobank.

Results

Automated analysis correlated highly with manual analysis for left and right ventricular volumes (all r > 0.95), strain (circumferential r = 0.89, longitudinal r > 0.89), and filling and ejection rates (all r ≥ 0.93). There was no significant bias for cardiac volumes and filling and ejection rates, except for right ventricular end-systolic volume (bias +1.80 ml; p = 0.01). The bias for FT strain was <1.3%. The sensitivity of detection of erroneous output was 95% for volume-derived parameters and 93% for FT strain. Finally, reference values were automatically derived from 2,029 CMR exams in healthy subjects.

Conclusions

The study demonstrates a DL-based framework for automated, quality-controlled characterization of cardiac function from cine CMR, without the need for direct clinician oversight.

Automatic quantification of the LV function and mass: A deep learning approach for cardiovascular MRI

OBJECTIVE

This paper proposes a novel approach for automatic left ventricle (LV) quantification using convolutional neural networks (CNN).

METHODS

The general framework consists of one CNN for detecting the LV, and another for tissue classification. Also, three new deep learning architectures were proposed for LV quantification. These new CNNs introduce the ideas of sparsity and depthwise separable convolution into the U-net architecture, as well as, a residual learning strategy level-to-level. To this end, we extend the classical U-net architecture and use the generalized Jaccard distance as optimization objective function.

RESULTS

The CNNs were trained and evaluated with 140 patients from two public cardiovascular magnetic resonance datasets (Sunnybrook and Cardiac Atlas Project) by using a 5-fold cross-validation strategy. Our results demonstrate a suitable accuracy for myocardial segmentation ( ∼ 0.9 Dice's coefficient), and a strong correlation with the most relevant physiological measures: 0.99 for end-diastolic and end-systolic volume, 0.97 for the left myocardial mass, 0.95 for the ejection fraction and 0.93 for the stroke volume and cardiac output.

CONCLUSION

Our simulation and clinical evaluation results demonstrate the capability and merits of the proposed CNN to estimate different structural and functional features such as LV mass and EF which are commonly used for both diagnosis and treatment of different pathologies.

SIGNIFICANCE

This paper suggests a new approach for automatic LV quantification based on deep learning where errors are comparable to the inter- and intra-operator ranges for manual contouring.

Monitoring Morphologic Changes in Male Rowers Using Limited Portable Echocardiography Performed by a Frontline Physician

Athletes' hearts have been studied for adaptive changes in response to exercise. Physiologic changes are normal responses to intense athletic training regimens, whereas pathologic changes predispose athletes to sudden cardiac death. The two alterations overlap in clinical presentation. Research continues to investigate the upper limits of cardiac remodeling to aid clinical decision making. Studying normal changes that occur in response to exercise will advance physicians' understanding of physiologic responses to exercise and potentially improve clinical distinction. To expand this body of knowledge, we present an observational case series that describes morphologic changes in athlete's hearts concurrent with performance measurements.

LV function validation of computer-assisted interventional system for cardiac resyncronisation therapy

PURPOSE

Cardiac resynchronisation therapy (CRT) is an established treatment for symptomatic patients with heart failure, a prolonged QRS duration, and impaired left ventricular (LV) function; however, non-response rates remain high. Recently proposed computer-assisted interventional platforms for CRT provide new routes to improving outcomes. Interventional systems must process information in an accurate, fast and highly automated way that is easy for the interventional cardiologists to use. In this paper, an interventional CRT platform is validated against two offline diagnostic tools to demonstrate that accurate information processing is possible in the time critical interventional setting.

METHODS

The study consisted of 3 healthy volunteers and 16 patients with heart failure and conventional criteria for CRT. Data analysis included the calculation of end-diastolic volume, end-systolic volume, stroke volume and ejection fraction; computation of global volume over the cardiac cycle as well as time to maximal contraction expressed as a percentage of the total cardiac cycle.

RESULTS

The results showed excellent correlation ([Formula: see text] values of [Formula: see text] and Pearson correlation coefficient of [Formula: see text]) with comparable offline diagnostic tools.

CONCLUSION

Results confirm that our interventional system has good accuracy in everyday clinical practice and can be of clinical utility in identification of CRT responders and LV function assessment.