EACVI survey on radiation exposure in interventional echocardiography

AIMS

The European Association of Cardiovascular Imaging (EACVI) Scientific Initiatives Committee performed a global survey on radiation exposure in interventional echocardiography. The survey aimed to collect data on local practices for radioprotection in interventional echocardiography and to assess the awareness of echocardiography operators about radiation-related risks.

METHODS AND RESULTS

A total of 258 interventional echocardiographers from 52 different countries (48% European) responded to the survey. One hundred twenty-two (47%) participants were women. Two-thirds (76%) of interventional echocardiographers worked in tertiary care/university hospitals. Interventional echocardiography was the main clinical activity for 34% of the survey participants. The median time spent in the cath-lab for the echocardiographic monitoring of structural heart procedures was 10 (5-20) hours/month. Despite this, only 28% of interventional echocardiographers received periodic training and certification in radioprotection and 72% of them did not know their annual radiation dose. The main adopted personal protection devices were lead aprons and thyroid collars (95% and 92% of use, respectively). Dedicated architectural protective shielding was not available for 33% of interventional echocardiographers. Nearly two-thirds of responders thought that the radiation exposure of interventional echocardiographers was higher than that of interventional cardiologists and 72% claimed for an improvement in the radioprotection measures.

CONCLUSION

Radioprotection measures for interventional echocardiographers are widely variable across centres. Radioprotection devices are often underused by interventional echocardiographers, portending an increased radiation-related risk. International scientific societies working in the field should collaborate to endorse radioprotection training, promote reliable radiation dose assessment, and support the adoption of radioprotection shielding dedicated to interventional echocardiographers.

Validation of 2D flow MRI for helical and vortical flows

PURPOSE

The main objective of this study was to develop two-dimensional (2D) phase contrast (PC) methods to quantify the helicity and vorticity of blood flow in the aortic root.

METHODS

This proof-of-concept study used four-dimensional (4D) flow cardiovascular MR (4D flow CMR) data of five healthy controls, five patients with heart failure with preserved ejection fraction and five patients with aortic stenosis (AS). A PC through-plane generated by 4D flow data was treated as a 2D PC plane and compared with the original 4D flow. Visual assessment of flow vectors was used to assess helicity and vorticity. We quantified flow displacement (FD), systolic flow reversal ratio (sFRR) and rotational angle (RA) using 2D PC.

RESULTS

For visual vortex flow presence near the inner curvature of the ascending aortic root on 4D flow CMR, sFRR demonstrated an area under the curve (AUC) of 0.955, p<0.001. A threshold of >8% for sFRR had a sensitivity of 82% and specificity of 100% for visual vortex presence. In addition, the average late systolic FD, a marker of flow eccentricity, also demonstrated an AUC of 0.909, p<0.001 for visual vortex flow. Manual systolic rotational flow angle change (ΔsRA) demonstrated excellent association with semiautomated ΔsRA (r=0.99, 95% CI 0.9907 to 0.999, p<0.001). In reproducibility testing, average systolic FD (FDsavg) showed a minimal bias at 1.28% with a high intraclass correlation coefficient (ICC=0.92). Similarly, sFRR had a minimal bias of 1.14% with an ICC of 0.96. ΔsRA demonstrated an acceptable bias of 5.72°-and an ICC of 0.99.

CONCLUSION

2D PC flow imaging can possibly quantify blood flow helicity (ΔRA) and vorticity (FRR). These imaging biomarkers of flow helicity and vorticity demonstrate high reproducibility for clinical adoption.

TRIALS REGISTRATION NUMBER

NCT05114785.

Concurrent left ventricular myocardial diffuse fibrosis and left atrial dysfunction strongly predicts incident heart failure and all-cause mortality

Funding Acknowledgements

Type of funding sources: Foundation. Main funding source(s): British Heart Foundation

Academy of Medical Sciences

Background

LV myocardial interstitial fibrosis has been reported to influence LA morphology and function via LV remodelling and diastolic dysfunction. However, this association, as well as their combined influence on clinical outcomes remains poorly characterised.

Aim

To evaluate the relationship between left ventricular (LV) fibrosis quantified by native T1 times and left atrial (LA) global and phasic function and their impact on clinical outcomes.

Methods

A total of 40,818 UK Biobank participants with cardiovascular magnetic resonance data were included. Native T1 mapping was performed using Shortened Modified Look-Locker Inversion recovery sequence with global myocardial T1 estimated by an automatic segmentation framework. Ten parameters of LA phasic function were calculated from normalised LA volume-time curves derived by a three-dimensional sparse active shape model. LV parameters (mass, end-diastolic volume, and ejection fraction) were extracted by a fully convolutional neural network. Multivariable regression models were used to assess the associations between T1 and LA parameters. Lastly, survival analysis was performed to assess the interplay between T1, LA function and incident heart failure, atrial fibrillation, major adverse cardiovascular event (MACE) and all-cause mortality.

Results

The mean age of study population was 64.0 ± 7.7 years; 47.8% were men. Higher T1 values were associated with larger LA minimum size (Beta= 0.89ml per 100ms; 95% confidence interval (CI) = 0.62, 1.17), and lower LA global emptying fraction (Beta= -0.012 per 100ms; CI= -0.015, -0.010), LA reservoir function (Beta= -0.060 per 100ms; CI= -0.083, -0.037) and LA booster function (Beta= -0.014 per 100ms; CI= -0.017, -0.011). Among LA phasic functional parameters, LA booster function is most strongly associated with T1. Survival analysis revealed concurrent high T1 and low LA function had a significant influence on incident heart failure (Hazard Ratio [HR] = 2.99; CI=1.91,2.01), atrial fibrillation (HR = 4.86; CI=3.51-6.54), MACE (HR = 1.86; CI = 1.36-2.54) and all-cause mortality (HR = 1.86; CI=1.22-2.82) compared to either parameter alone, even after accounting for LV parameters (Figure 1).

Conclusion

This is the first study to robustly demonstrate the associations between myocardial diffuse fibrosis and reduced LA global and phasic functional measurements. We reveal the independent prognostic role of high T1 values accompanied by low LA function in predicting adverse clinical outcomes in a general population. These findings advance our understanding of the relationships between myocardial fibrosis and LA biomechanics at an early, subclinical stage, and highlight the additive value of incorporating these biomarkers into clinical decision making.

Influence of novel risk markers on defibrillator implantation in hypertrophic cardiomyopathy

Background/Introduction

To guide implantable defibrillator (ICD) use in hypertrophic cardiomyopathy (HCM), European Society of Cardiology (ESC) guidelines recommend using individualised sudden death (SCD) risk scores based on quantitative clinical data. Newly published American guidelines are based on the accumulation of binary risk markers, which include imaging-based novel risk markers (NRMs) that were absent from the development of the ESC's algorithm. These NRMs are ejection fraction &lt;50%, apical aneurysm, and extensive late gadolinium enhancement on cardiac magnetic resonance (CMR) imaging.

Purpose

To assess how NRMs may have altered ICD prescription across ESC-based SCD risk status prior to publication of current American guidance.

Methods

We examined electronic records (2013–2020) of a subset of HCM patients with contemporaneous (within 12 months) CMR and echocardiography data for NRMs, ESC risk status, and ICD prescription. Differences in categorical data were assessed by Fisher's exact test.

Results

We studied 334 HCM patients (74% male; age: 58±14 years), of whom 83 (25%) were referred for ICD. ESC risk status was considered low, medium (4–6% 5-year SCD risk), or high in 264, 26, and 20 patients, for whom ICDs were recommended in 40 (15%), 20 (77%), and 18 (90%) patients, respectively. In patients with low SCD risk status, rate of ICD recommendation was significantly higher when ≥1 NRMs were present (34/126 – 27% vs. 0 NRMs: 6/138 – 4%; p&lt;0.0001). NRMs did not appear to influence ICD recommendation in patients with medium (≥1 NRMs: 14/17 – 82% vs. 0 NRMs: 6/9 – 67%; p=0.6) or high (≥1 NRMs: 14/15 – 93% vs. 0 NRMs: 4/5 – 80%; p=0.4) SCD risk status (Figure 1). NRMs were less frequent in low risk patients than in high risk patients (126/264 – 48% vs. 15/20 – 75%; p&lt;0.05), suggesting interaction between ESC status and NRMs (Figure 2).

Conclusion

NRMs have disproportionate influence on ICD prescription in low ESC risk HCM patients. However, NRMs are not independent of ESC risk status, suggesting iterative development of the ESC's algorithmic approach will be the most effective way of predicting SCD.

Funding Acknowledgement

Type of funding sources: None.

Cardiac magnetic resonance radiomics for prediction of incident heart failure: a feasibility study in the UK Biobank Imaging cohort

Background

Cardiac magnetic resonance (CMR) radiomics is a novel image quantification technique with the potential to improve image-based disease diagnosis and prediction.

Purpose

In this proof-of-concept study, we aimed to evaluate the utility of CMR radiomics in the prediction of incident heart failure (HF).

Methods

We studied 32,121 UK Biobank participants with CMR. Incident HF was defined from linked Hospital Episode Statistics. To create a balanced cohort, we identified as comparators an equal number of randomly selected subjects who did not develop the outcome of interest during this period. Radiomics shape, first-order and texture features were extracted from short-axis cine images (left and right ventricle, left ventricular myocardium) using the Pyradiomics toolbox. Vascular risk factors (VRFs) were considered as additional predictors. Feature selection was conducted using the sequential forward selection technique and modelling was performed using Support Vector Machine (SVM) methods with 5-fold cross-validation. Models were developed using 1) VRFs alone, 2) radiomics alone, and 3) VRFs and radiomics. We determined model performance using receiver operating characteristic (ROC) curve and area under the curve (AUC) scores.

Results

Over average follow-up time of 3.7 (±1.3) years, 209 participants experienced incident HF. Among vascular risk factors, age, body size, hypertension, diabetes, high cholesterol were chosen for the incident HF predictive model (Accuracy: 0.66, AUC: 0.73) by the SVM methods. The model based on radiomics features reached a marginal improvement compared to vascular risk factors alone (Accuracy: 0.71, AUC: 0.75). The combination of VRFs and radiomics features significantly improved the performance of the model to predict incident HF compared to VRFs alone (Accuracy: 0.77; AUC: 0.83; p&lt;0.05)

Conclusion

We demonstrate the feasibility of CMR radomics features to predict incident HF and illustrate their added value over vascular risk factors.

Funding Acknowledgement

Type of funding sources: Public grant(s) – EU funding.

The association between native myocardial T1 relaxation times and left atrial phasic structure and function: the UK Biobank Imaging Enhancement study

Introduction

Left ventricular (LV) myocardial fibrosis is posited to result in left atrial (LA) changes via LV remodelling and diastolic dysfunction, though the association remains poorly characterised. Native myocardial T1 mapping is a non-invasive modality that quantifies diffuse myocardial fibrosis. This study examines the relationship between LV fibrosis (quantified by native T1 times) and LA function, drawing upon data from the UK Biobank.

Methods

40,818 participants underwent cardiovascular magnetic resonance (CMR) using steady-state free precession imaging at 1.5 Tesla. Native T1-mapping was performed using the Shortened Modified Look-Locker Inversion recovery technique (ShMOLLI), with global myocardial T1 estimated by an automatic segmentation framework. Nine parameters of LA phasic function were calculated (representing global, reservoir, conduit and booster components) from normalised LA volume-time curves. LV parameters (LV Mass, end-diastolic volume and ejection fraction) were extracted by a convolutional neural network. Multivariable logistic regression models were used to assess the association between T1 (exposure) and LA function (outcome). Mediation analysis was performed to assess the role of LV parameters as a mediator for the association between T1 and LA function. Lastly, potential non-linear relationships between T1 and LA function were investigated using Restrictive Cubic Spline (RCS) modelling, with model fit assessed via the Akaike Information Criterion (AIC).

Results

Higher T1 values were positively associated with larger LA volumes, and negatively associated with markers of LA global, reservoir and booster function. In the fully adjusted model, T1 was positively associated with larger LA minimum size (Beta: +0.034 SD per T1 SD; Confidence Interval (CI): 0.024, 0.045), and negatively associated with LA emptying volume (Beta: −0.017; CI: −0.027, −0.006), LA booster volume (Beta: −0.019; CI: −0.030, −0.008), LA emptying fraction (Beta: −0.052; CI: −0.062, −0.041), and LA reservoir function (Beta: −0.028; CI: −0.039, −0.017). Though adjustment for LV parameters did not fully attenuate the above relationships, LV parameters were consistent mediators between T1 and LA function, with proportional mediative effects ranging from 15% to 75%. Lastly, there is evidence of an inverted J-shaped relationship between T1 and LA function, with the associations becoming more apparent in the upper half of T1 ranges (turning points within 925–950 ms, median T1 = 930 ms) (p&lt;0.05).

Conclusion

This study demonstrates a consistent association between higher native T1 values (as a marker of myocardial fibrosis) and lower LA global and phasic functions. We also highlighted an interplay between T1 values, LV remodelling and LA dysfunction. These findings will facilitate our understanding of the disease processes underlying cardiac dysfunction and myocardial remodelling at an early, subclinical stage.

Funding Acknowledgement

Type of funding sources: Public hospital(s). Main funding source(s): This work was part of the portfolio of translational research of the National Institute for Health Research Biomedical Research Centre at Barts and The London School of Medicine and DentistryDr Nay Aung is supported by a Wellcome Trust Research Training Fellowship (203553/Z/16/Z)

Patterns and outcomes of cardiovascular emergency department encounters for men and women in the USA

Aims

We described sex-differential disease patterns and outcomes of over 20.6 million cardiovascular emergency department (ED) encounters in the USA.

Methods and results

We analysed primary cardiovascular encounters from the Nationwide Emergency Department Sample (between 2016–2018). We grouped the documented cardiovascular diagnoses into 15 disease categories. The sample included 48.7% women; average age was 67 [54,78] years. Men had greater overall baseline co-morbidity burden; however, women had higher rates of obesity, hypertension, and cerebrovascular disease. For women, the most common ED encounters were essential hypertension (16.0%), hypertensive heart or kidney disease (14.1%), and atrial fibrillation (AF)/flutter (10.2%). For men, the most common encounters were hypertensive heart or kidney disease (14.7%), essential hypertension (10.8%), and acute myocardial infarction (AMI, 10.7%). Women were more likely to present with essential hypertension, hypertensive crisis, AF/flutter, supraventricular tachycardia, pulmonary embolism, or ischaemic stroke. Men were more likely to present with AMI or cardiac arrest. In logistic regression models adjusted for baseline covariates, women with intracranial haemorrhage had higher risk of hospitalisation and death. Women with ischaemic stroke had higher risk of hospitalisation and death in ED. Women presenting with pulmonary embolism were less likely to be hospitalised but were more likely to die. Women with aortic aneurysm/dissection had higher risk of hospitalisation and death. Men were more likely to die following presentations with hypertensive heart or kidney disease, AF/flutter, AMI, or cardiac arrest.

Conclusion

In this large nationally representative sample of cardiovascular ED presentations, we demonstrate significant sex differences in disease distribution, hospitalisation, and death.

Funding Acknowledgement

Type of funding sources: None.

Cardiac magnetic resonance imaging predictors of ventricular arrhythmia in mid-cavity obstructive hypertrophic cardiomyopathy

Background/Introduction

Left ventricular (LV) mid-cavity obstruction (LVMCO) in hypertrophic cardiomyopathy (HCM) is an uncommon phenotypic feature predisposed to the formation of myocardial fibrosis and apical aneurysms (LVAA). These features may be independently proarrhythmic, and LVAA is considered a class 2a indication for implantable cardioverter defibrillator (ICD) in current US, but not European guidelines for the primary prevention of sudden cardiac death (SCD). Cardiac magnetic resonance (CMR) imaging is the preferred modality for detecting these and other phenotypic features critical to SCD risk assessment.

Purpose

To assess the ability of CMR imaging parameters to predict occurrence of non-sustained ventricular tachycardia (NSVT) in HCM patients with Doppler-derived evidence of LVMCO.

Methods

Multi-modality imaging records were retrospectively assessed to identify HCM patients with Doppler-LVMCO and CMR scans. CMR images were assessed by an investigator blinded to clinical status. Late gadolinium enhancement (LGE) was quantified using the full-width, half-maximum technique. CMR imaging parameters were assessed for predictive ability using Cox proportional hazards during univariate and multivariate analyses, accounting for time to event (NSVT or censorship of follow-up).

Results

The study cohort included 58 patients (57±11 years, 74% male) with a median follow-up of 6.2 (IQR 4.3) years. Mean mid-cavity gradient was 33±23 mmHg. NSVT was detected in 27/58 (47%) patients, was 4 beats or longer in 23/27 (85%) and was monomorphic in 21/27 (77%).

On univariate analysis, predictors of NSVT during follow-up include LV mass index (HR 1.02, 95% CI 1.00–1.04, p=0.03), LGE in grams (HR 1.04, 95% CI 1.01–1.06, p=0.005), and LVAA (HR 2.57, 95% CI 1.14–5.79, p=0.023). After multivariate adjustment (Table 2), none were significantly associated.

Conclusions

In LVMCO, magnitude of LV hypertrophy, extent of LGE and the presence of an apical aneurysm may not be independent predictors of ventricular arrhythmias. SCD algorithms based on qualitative assessments of these features may overestimate risk.

Funding Acknowledgement

Type of funding sources: Public Institution(s). Main funding source(s): National Institute of Health Research (NIHR)

Diagnostic accuracy of stress CMR to evaluate chronic coronary syndromes: an updated meta-analysis

Background

Stress cardiovascular magnetic resonance (CMR) imaging is utilised for the evaluation of patients with stable chest pain and intermediate or high pre-test likelihood of coronary artery disease (CAD).

Purpose

To provide an updated synthesis of diagnostic accuracy of stress CMR imaging for the diagnosis of anatomically and functionally significant CAD in patients with stable chest pain and suspected or known CAD.

Methods

After prospective registration and approval of the study protocol, we performed a systematic review and meta-analysis of studies published between 2000 through 2021, enrolling ≥100 patients, and reporting on the diagnostic accuracy of stress CMR imaging to diagnose anatomically and functionally significant CAD with invasive coronary angiography (ICA) or ICA and fractional flow reserve (FFR &lt;0.80) as the reference standard. The novel split component synthesis method was used through the SCSmeta function in R. The meta-analysis yielded pooled diagnostic indicators including diagnostic odds ratio (DOR), sensitivity, specificity, positive likelihood ratio (pLR), negative likelihood ratio (nLR) and area under the curve (AUC).

Results

We identified a total of 32 studies pooling an overall population of 7,678 individuals (mean age 62 years, 70% males, CAD prevalence 52%). Compared with ICA (29 studies, 7,360 patients), stress CMR yielded a pooled DOR of 19.2 (95% CI: 12.5–29.4), a sensitivity of 84% (95% CI: 79–88%), a specificity of 79% (95% CI: 73–84%), a pLR of 3.9 (95% CI: 3.0–5.3), a nLR of 0.2 (95% CI: 0.2–0.3), and an AUC of 0.81 (95% CI: 0.78–0.84) for the detection of anatomically obstructive CAD. Compared with ICA and FFR (8 studies, 1,196 patients), stress CMR yielded a pooled DOR of 26.4 (95% CI: 10.6–65.9), a sensitivity of 81% (95% CI: 68–89), a specificity of 86% (95% CI: 75–93%), a pLR of 5.8 (95% CI: 3.0–11.4), a nLR of 0.2 (95% CI: 0.1–0.4), and an AUC of 0.84 (0.77–0.89) for the detection of functionally obstructive CAD. Higher diagnostic accuracy was observed for 3 Tesla myocardial perfusion imaging studies to detect both anatomically and functionally obstructive CAD, with pooled DORs of 24.3 and 33.2, respectively.

Conclusions

In patients with stable chest pain and known or suspected CAD, stress CMR imaging yields high diagnostic accuracy to detect both anatomically and functionally obstructive CAD. Stress CMR perfusion imaging at 3 Tesla is to be associated with overall greater diagnostic accuracy.

Funding Acknowledgement

Type of funding sources: None.

Prognostic yield of stress CMR to evaluate chronic coronary syndromes: a systematic review and meta-analysis

Background

Assessment of ischemia with stress cardiovascular magnetic resonance (CMR) imaging is recommended in patients with stable chest pain and intermediate or high pre-test probability of coronary artery disease (CAD).

Purpose

To provide an updated synthesis on prognostic significance of stress CMR imaging in patients with stable chest pain and suspected or known CAD.

Methods

After prospective registration and approval of the study protocol, we performed a systematic review and meta-analysis of studies published between 2000 through 2021, enrolling ≥100 patients, and reporting outcome data of CAD patients undergoing stress CMR. Odds ratios (ORs) and 95% confidence intervals (CIs) for all-cause death, cardiovascular (CV) death and major adverse cardiac events (MACE: CV death and myocardial infarction), were pooled through inverse variance random-effects meta-analysis to compute summary effect size. Annualized event rates (AERs) were extracted from each study and compared by χ2-statistic. A warranty period was defined as the time interval with an AER &lt;1%.

Results

We identified a total of 33 studies pooling an overall tested population of 68920 patients (mean age 62 years; 56% males; known CAD 32%; 386117 person-years). Ischemia was found in 13617 (20%). Mean follow-up was 3.5±2.1 years. Presence of ischemia was associated with increased risk of all-cause death (OR 2.0 95% CI: 1.7–2.3), CV death (OR 6.4 95% CI: 4.5–9.1), and MACE (OR 5.0 95% CI: 3.6–6.8). Cumulative AERs for all-cause death, CV death and MACE were 2.97%, 2.51%, and 3.99% in patients with ischemia, and 1.40%, 0.59%, and 0.98% in patients without ischemia, respectively (p&lt;0.0001 for all comparisons).

Conclusion

Stress CMR imaging yields robust prognostic information in patients with suspected or known CAD. Presence of ischemia is associated with increased risk of all-cause death, CV death and MACE. Patients with negative stress CMR have a very low risk (&lt;1%) of CV death and MACE with a warranty period of at least 3.5 years.

Funding Acknowledgement

Type of funding sources: None.

Annotation and quality assessment of left ventricular filling and relaxation pattern using one-dimensional convolutional neural network

Funding Acknowledgements

Type of funding sources: Public grant(s) – National budget only. Main funding source(s): National Institute For Health Research (NIHR), UK

Introduction

Aberrations in left ventricular (LV) filling or relaxation – known as diastolic dysfunction – occur in heart failure with preserved ejection fraction. CMR is the reference modality for the assessment of ventricular systolic function, however, its role in evaluation of diastolic function is limited at present. One promising technique to assess diastolic function by CMR is the derivation of LV filling and emptying rates from the volume-time curves of cine images.

Purpose

To automatically assess the quality of LV filling-rate curves and annotate the peak emptying and filling rates.

Methods

A previously-described deep-learning network was used to automatically segment the entire cardiac cycle captured by short-axis SSFP cine images from the UK Biobank1. The LV filling-rate curves derived from the volume-time data were smoothed with Savitzky–Golay filter. The peak emptying rate (PER), early peak filling rate (PFR-E) and late peak filling rate (PFR-A) were first annotated by a simple peak finding algorithm from Python Scipy signal module. The preliminary annotated curves were reviewed by five human experts (i) to check for peak-annotation errors and (ii) to provide the curve quality score ranging from 1 to 3 for each peak (score 1 denotes good quality, score 2 represents moderate quality and score 3 indicates poor quality). Higher total score (minimum = 3, maximum = 9), therefore, represents poorer overall curve quality. This expert-annotated dataset was used to train two separate one-dimensional convolutional neural networks (1D-CNN) (Figure 1) for peak annotation and curve quality assessment (QA) using Tensorflow library in Python.

Results

The data from 6,328 LV filling-rate curves were split into the training and testing sets (80:20). The fine-tuned 1D-CNN comprising six hidden layers with two residual connections annotated the PER, PFR-E and PFR-A with the test-set accuracy of 95%, 95% and 98%, respectively. A second trained 1D-CNN for QA based on similar architecture predicted the overall curve quality score with a small error rate (mean absolute error: 0.46, mean squared error: 0.68). These two networks were used to quality check and label 19,409 UK Biobank CMR studies (See Figure 2 for exemplary results). After removing data from poor-quality curves (quality score ≥ 5), 18,735 studies remained. The mean±standard deviation of PER, PFR-E and PFR-A are 461±110 ml/s, 359±117 ml/s and 336±120 ml/s, respectively. Ageing is associated with lower PFR-E (−58.4 ml/s, 95% confidence interval [CI]: −56.1 to −60.7 ml/s per decade increment) and higher PFR-A (18.3 ml/s, 95% CI: 15.8 to 20.8 ml/s per decade increment).

Conclusion

The 1D-CNN models can be used to automatically grade the quality of LV filling rate curves and label important diastolic parameters with a high level of accuracy. The derived data recapitulate impaired LV relaxation pattern associated with ageing and can be used as surrogate indices of diastology by CMR. Figure 1Figure 2

AlzEye: longitudinal record-level linkage of ophthalmic imaging and hospital admissions of 353 157 patients in London, UK

PURPOSE

Retinal signatures of systemic disease ('oculomics') are increasingly being revealed through a combination of high-resolution ophthalmic imaging and sophisticated modelling strategies. Progress is currently limited not mainly by technical issues, but by the lack of large labelled datasets, a sine qua non for deep learning. Such data are derived from prospective epidemiological studies, in which retinal imaging is typically unimodal, cross-sectional, of modest number and relates to cohorts, which are not enriched with subpopulations of interest, such as those with systemic disease. We thus linked longitudinal multimodal retinal imaging from routinely collected National Health Service (NHS) data with systemic disease data from hospital admissions using a privacy-by-design third-party linkage approach.

PARTICIPANTS

Between 1 January 2008 and 1 April 2018, 353 157 participants aged 40 years or older, who attended Moorfields Eye Hospital NHS Foundation Trust, a tertiary ophthalmic institution incorporating a principal central site, four district hubs and five satellite clinics in and around London, UK serving a catchment population of approximately six million people.

FINDINGS TO DATE

Among the 353 157 individuals, 186 651 had a total of 1 337 711 Hospital Episode Statistics admitted patient care episodes. Systemic diagnoses recorded at these episodes include 12 022 patients with myocardial infarction, 11 735 with all-cause stroke and 13 363 with all-cause dementia. A total of 6 261 931 retinal images of seven different modalities and across three manufacturers were acquired from 1 54 830 patients. The majority of retinal images were retinal photographs (n=1 874 175) followed by optical coherence tomography (n=1 567 358).

FUTURE PLANS

AlzEye combines the world's largest single institution retinal imaging database with nationally collected systemic data to create an exceptional large-scale, enriched cohort that reflects the diversity of the population served. First analyses will address cardiovascular diseases and dementia, with a view to identifying hidden retinal signatures that may lead to earlier detection and risk management of these life-threatening conditions.

A Systematic Quality Scoring Analysis to Assess Automated Cardiovascular Magnetic Resonance Segmentation Algorithms

Background

The quantitative measures used to assess the performance of automated methods often do not reflect the clinical acceptability of contouring. A quality-based assessment of automated cardiac magnetic resonance (CMR) segmentation more relevant to clinical practice is therefore needed.

Objective

We propose a new method for assessing the quality of machine learning (ML) outputs. We evaluate the clinical utility of the proposed method as it is employed to systematically analyse the quality of an automated contouring algorithm.

Methods

A dataset of short-axis (SAX) cine CMR images from a clinically heterogeneous population (n = 217) were manually contoured by a team of experienced investigators. On the same images we derived automated contours using a ML algorithm. A contour quality scoring application randomly presented manual and automated contours to four blinded clinicians, who were asked to assign a quality score from a predefined rubric. Firstly, we analyzed the distribution of quality scores between the two contouring methods across all clinicians. Secondly, we analyzed the interobserver reliability between the raters. Finally, we examined whether there was a variation in scores based on the type of contour, SAX slice level, and underlying disease.

Results

The overall distribution of scores between the two methods was significantly different, with automated contours scoring better than the manual (OR (95% CI) = 1.17 (1.07–1.28), p = 0.001; n = 9401). There was substantial scoring agreement between raters for each contouring method independently, albeit it was significantly better for automated segmentation (automated: AC2 = 0.940, 95% CI, 0.937–0.943 vs manual: AC2 = 0.934, 95% CI, 0.931–0.937; p = 0.006). Next, the analysis of quality scores based on different factors was performed. Our approach helped identify trends patterns of lower segmentation quality as observed for left ventricle epicardial and basal contours with both methods. Similarly, significant differences in quality between the two methods were also found in dilated cardiomyopathy and hypertension.

Conclusions

Our results confirm the ability of our systematic scoring analysis to determine the clinical acceptability of automated contours. This approach focused on the contours' clinical utility could ultimately improve clinicians' confidence in artificial intelligence and its acceptability in the clinical workflow.

Multi-modality imaging in hypertrophic cardiomyopathy: intermodal discrepancies in key prognostic parameters

Background/Introduction

Multi-modality imaging is crucial for confirming diagnosis and assessing prognosis in patients with hypertrophic cardiomyopathy (HCM). However, inter-modality discrepancies in key parameters are commonly reported.

Purpose

To assess real-world inter-modal reporting discrepancies between transthoracic echocardiography (TTE) and cardiac magnetic resonance (CMR) imaging in the measurement of four key parameters in HCM patients.

Methods

Consecutive HCM patients with TTE and CMR performed within 6 months of each other at a tertiary centre were retrospectively assessed for reported maximum wall thickness (MWT), left atrial diameter (LAd), left ventricular ejection fraction (LVEF) and presence of left ventricular apical aneurysm (LVAA). The CMR report was considered gold standard. Data are reported as mean ± standard deviation (SD) or median and interquartile range (IQR) as appropriate.

Results

353 consecutive HCM patients (72% male, median age 60.9 years, IQR 49.8–71.6 years) with TTE and CMR within 6 months (median difference 1.7 months, IQR 1.1–3.4 months) were assessed between 4th January 2018 and 9th April 2019. Of 284 patients with paired MWT data, median difference was 0.0 mm (IQR −1.0 to 3.0 mm, p=0.02), likely representing a difference in distributions of MWT. TTE both over and underestimated MWT (in 36% and 46% cases respectively).

Of the 94 patients with paired LAd data, mean difference was 0.4±5.7 mm (95% CI −0.8010 to 1.546, p=0.5).

N=320 patients with paired LVEF data (after excluding patients with atrial fibrillation (n=20)). Median difference in LVEF was 12% (IQR 5–19% p&lt;0.0001). TTE underestimated LVEF in 88% of cases. CMR and TTE both identified 14 (5%) patients as having LVEF &lt;50%. There were however 8 cases of disagreement in classification of LVEF &lt;50%, due to over (n=4) or underestimation (n=4) by TTE.

LVAA was accurately identified by TTE in only 9/30 (30%) of those patients with demonstrable LVAA by CMR (p=0.0008). TTE evidence of a discreet apical chamber (paradoxical jet on spectral or colour Doppler) was present in 16/21 (76%) cases where TTE failed to overtly identify LVAA. However, apical or mid-cavity obliteration was reported in 15/21 (71%) cases where TTE failed to identify LVAA.

Conclusion(s)

Echocardiography and CMR measurements are often used interchangeably in clinical practice but inter-modality discrepancies can affect diagnosis and sudden cardiac death (SCD) risk assessment. This is particularly important for binary risk factors such as LVEF&lt;50% or LVAA which are considered major SCD risk factors in the latest American Heart Association guidelines. 25 (7%) patients in our cohort had major risk factors identified by CMR that were not identified on TTE. CMR is an important, recommended tool where TTE imaging is suboptimal, but attention to more subtle elements of abnormal intracavity blood flow may be able to increase LVAA detection during TTE.

Funding Acknowledgement

Type of funding sources: None.

Association of daily coffee consumption with cardiovascular health – results from the UK Biobank

Background

There are conflicting reports on the association of coffee consumption with cardiovascular (CV) health. The UK Biobank is a prospective cohort study including data for half a million middle-aged individuals.

Purpose

We studied the association of daily coffee consumption with all-cause and CV mortality, and incidence of the major CV diseases in the UK Biobank. In a subgroup of participants who underwent cardiovascular magnetic resonance (CMR), we evaluated the association between regular coffee intake and cardiac structure and function parameters.

Methods

UK Biobank cohort of participants without clinically manifested heart disease at the time of recruitment were included. Regular coffee intake was categorized into 3 groups: zero, light-to-moderate (0.5–3 cups/day) and high (&gt;3 cups/day) coffee drinkers. We estimated association of daily coffee consumption with incident outcomes using multivariable Cox-regression models (median follow-up of 11 years) and, in the subset with CMR data, with left and right ventricular (LV, RV) end-systolic and end-diastolic volumes, LV mass, and LV/RV stroke volume using multivariable linear regression. Models were adjusted for potential confounders and mediators, including: age, sex, non-European ethnicities, body mass index, smoking, physical activity, Townsend deprivation index, alcohol, meat, fruit and vegetable intake, hypertension, diabetes mellitus, and cholesterol level.

Results

We included 468,629 individuals (mean age 56.2±8.1 years, 44.2% male). Among them, 22.1% did not consume coffee on a regular basis, 58.4% had 0.5–3 cups per day and 19.5% had &gt;3 cups per day. After adjustment for potential confounders and mediators, compared to non-coffee drinkers, light-to-moderate coffee drinking was associated with lower risk of all-cause mortality (HR=0.88, p&lt;0.001), CV mortality (HR=0.83, p=0.006), and incident stroke (HR=0.79; p=0.037). CMR data were available in 30,650 participants. In multivariable analysis, compared to non-coffee drinkers, both the light-to-moderate and high coffee consuming categories, were associated with significantly increased LV and RV ventricular end-systolic (β=0.91 and 1.64 for LV and 1.10 and 1.72 for RV), end-diastolic (β=2.21 and 3.28 for LV and 2.24 and 3.35 for RV) and stroke volumes (β=1.31 and 1.64 for LV and 1.15 and 1.63 for RV), as well as greater LV mass (β=0.78 and 1.64; all p&lt;0.001).

Conclusion

In this large study of the UK Biobank population, regular coffee consumption of up to 3 cups per day was associated with favorable cardiovascular outcomes, in particular, decreased all-cause and CV mortality and stroke incidence. Regular coffee consumption was also associated with a pattern of CMR metrics in keeping with the reverse of age-related cardiac alterations.

Funding Acknowledgement

Type of funding sources: None.

Fairness in AI: are deep learning-based CMR segmentation algorithms biased?

Background/Introduction

Artificial intelligence (AI) is providing opportunities to transform cardiovascular medicine. A particular challenge in the application of AI technology is their potential for intrinsic and extrinsic biases, such as those based on gender and/or ethnicity. Unless satisfactorily addressed, these biases could lead to inequalities in early diagnosis, treatments and outcomes. Fairness in AI is a relatively new but fast-growing research field which deals with assessing and addressing potential bias in AI models.

Purpose

To perform the first analysis that assesses bias in AI-based cardiac MR segmentation models in a large-scale database.

Methods

A state-of-the-art deep learning (DL) based segmentation network, the “nnU-Net” framework [1], was used for automatic segmentation of both ventricles and the myocardium from cine short-axis cardiac MR over the full cardiac cycle. The dataset used consisted of end-diastole and end-systole short-axis cine cardiac MR images of 5,903 subjects (61.5±7.1 years). The nnU-Net network was trained and evaluated using a 5-fold cross validation (splits: train 60% / validation 20% / test 20%). Data on race and gender were obtained from the UK Biobank database and their distribution is summarized in Figure 1. To assess gender and racial bias in the segmentation network, we compared the Dice scores - which measure the overlap between manual and automatic segmentations – and the absolute error in measurements of biventricular volumes and function between patients grouped by ethnicity and gender.

Results

Figure 2 shows the Dice scores and the volumetric and functional measures for the full database, stratified by gender and by ethnicity. Results on the overall population showed an excellent agreement between the manual and automatic segmentations which is consistent with previous reported results [2–3]. However, we find statistically significant differences in Dice scores as well as volumetric measures between different ethnicities, showing that the segmentation network is biased against minority racial groups. No significant differences were found in Dice scores between genders. Similarly, for the end diastolic, end systolic volumes and ejection fraction, there were statistically significant differences in absolute error between the overall population and all racial groups except white.

Conclusion(s)

We have shown, for the first time, that racial bias exists in DL-based cardiac MR segmentation models. Our hypothesis is that this bias is a result of the unbalanced nature of the training data, and this is supported by the results which show that there is racial bias but not gender bias when trained using the UK Biobank database, which is gender-balanced but not race-balanced. In this work we want to highlight the potential issue of bias in DL-based image segmentation models when translating into a clinical environment.

Funding Acknowledgement

Type of funding sources: Public grant(s) – National budget only. Main funding source(s): - EPSRC- Wellcome EPSRC Centre for Medical Engineering at the School of Biomedical Engineering and Imaging Sciences, King's College London Figure 1Figure 2

Cost-Minimization analysis for cardiac revascularization in 12 healthcare systems based on the EuroCMR/SPINS registries

Funding Acknowledgements

Type of funding sources: Private company. Main funding source(s): EuroCMR registry (Life Sciences GE Healthcare, Medtronic Inc., Minneapolis MN, USA; Novartis International AG, Basel, Switzerland; Siemens Healthcare, Erlangen, Germany), SPINS registry (Siemens Healthineers, Erlangen, Germany; Bayer AG, Leverkusen, Germany)

Background

Coronary artery disease (CAD) is a major contributor to the public health burden. Stress perfusion cardiac magnetic resonance (CMR) has an excellent accuracy to detect CAD, but data on its cost effectiveness are scarce.

Purpose

To compare the costs of a CMR-guided strategy vs 2 invasive strategies based on 2 large international CMR registries.

Methods

In the EuroCMR registry (n = 3’647, 59 centers, 18 countries) and the US-based SPINS registry (n = 2’349, 13 centers, 11 states) costs were calculated for 12 healthcare systems (8 Europe, US, 2 Latin America, 1 Asia). They included diagnostic examinations (CMR, X-ray coronarography (CXA) with/without FFR), revascularizations, and complications during a 1-year follow-up. Endpoints in both registries were all-cause and cardiovascular (CV) death, sudden cardiac death (SCD), aborted SCD, non-fatal myocardial infarction (nf-MI), and stroke. 7 sub-group analyses covered low to high-risk cohorts. Patients with ischemia-positive CMR underwent CXA and revascularization (percuteneous and surgical intervention) at the treating physician’s discretion (=CMR + CXA-strategy). In the hypothetical invasive CXA + FFR-strategy, costs were calculated for an initial CXA and an FFR in vessels with ≥50% stenoses assuming the same proportion of revascularizations/complications as in the CMR + CXA-strategy and FFR positive rates as given in the literature. In the CXA-only strategy, costs included CXA and revascularizations of ≥50% stenoses.

Results

Revascularizations were performed in 8.0% and 6.2% (p &lt; 0.01) of SPINS and EuroCMR patients, respectively. Consistent cost savings were observed for the CMR + CXA strategy vs CXA + FFR in all 12 healthcare systems ranging from 42 ± 20% and 52 ± 15% in the low-risk EuroCMR and SPINS patients with atypical chest pain (CV-death and nf-MI 0.4-0.7%/y), respectively, to 31 ± 16% in the high-risk SPINS patients (CV-death and nf-MI 3.2%/y) with known CAD (p &lt; 0.0001 vs 0 in all groups, Fig 1/2). Cost savings were even higher vs CXA-only with 63 ± 11%, 73 ± 6%, and 52 ± 9%, respectively (p &lt; 0.0001 vs 0 in all groups, Fig 2).

Conclusions

In 12 healthcare systems, a CMR + CXA-strategy yielded consistent moderate to high cost savings compared to a hypothetical CXA + FFR-strategy over the entire spectrum of risk. Cost savings were consistently high vs a CXA-only strategy for all risk groups.

Figure 1: SPINS refers to the subgroup of patients with suspected CAD (n = 1’530), EuroCMR (= suspected CAD; n = 3’647). EuroCMR vs SPINS ns. Countries per region are listed in alphabetical order.

Figure 2: Top: CMR + CXA vs CXA + FFR: ANOVA: overall p = 0.0017, * vs EuroCMR typ angina: p &lt; 0.005 (Scheffe post-hoc testing). Bottom: CMR + CXA vs CXA-only: ANOVA overall p &lt; 0.0001, * vs SPINS with CAD and vs EuroCMR typ A: p &lt; 0.0001; † vs SPINS with CAD: p &lt; 0.03; ‡ vs EuroCMR typ A: p &lt; 0.0001; § vs SPINS with CAD: p &lt; 0.002; ║ vs EuroCMR typ: p &lt; 0.002 (Scheffe post-hoc tesing)

Automated myocardial segmentation in native t1-mapping cardiovascular magnetic resonance images based on machine learning: a validation study in the UK biobank"s covid-19 subset

Funding Acknowledgements

Type of funding sources: Public grant(s) – National budget only. Main funding source(s): Innovate UK

Background

 Regional assessment of septal native T1 values with cardiovascular magnetic resonance (CMR) is used to characterise diffuse myocardial diseases. Previous studies suggest its potential role in detecting early pathological alterations, which may help identify high-risk subjects at early disease stages. Automated analysis of myocardial native T1 images may enable faster CMR analysis and reduce inter-observer variability of manual analysis. However, the technical performance of such methodologies has not been previously reported.

Purpose

 We tested, in a subset of UK Biobank participants, the degree of agreement between CMR septal myocardial T1 values obtained from our machine learning (ML) algorithm and septal native T1 values computed from manual segmentations.

Methods

 We analysed the first 292 participants who were tested for severe acute respiratory syndrome coronavirus 2 (SARS-CoV-2) and had CMR imaging (1.5 Tesla, Siemens MAGNETOM Aera). T1 mapping was performed in a single mid-ventricular short axis (SAX) slice using ShMOLLI (WIP780B) sequences. Three experienced CMR readers independently measured native T1 values by manually placing a single region of interest (ROI) covering half of the anteroseptal and half of the inferoseptal wall using cvi42 post-processing software (version 5.11). A mean T1 value for each participant was then calculated. A ML algorithm developed by Circle Cardiovascular Imaging Inc. was then applied to the same images to derive the myocardium T1 values automatically. The algorithm was previously trained to segment myocardium from SAX T1 and non-T1 mapping images on two external CMR datasets. We compared the mean septal ROI T1 values to the mean myocardium T1 values predicted by the ML algorithm.

Results

 Two studies were excluded after quality control. The ML-derived and the manually calculated mean T1 values were significantly correlated (r = 0.82, p < 0.001). The Bland-Altman analysis between the two methods showed a mean bias of 3.64 ms, with 95% limits of agreement of −38.88 to 53.46 ms, indicating good agreement (figure 1).

Conclusions

 We demonstrated strong correlation and good agreement between native T1 values obtained from our automated analysis method and manual T1 septal analysis in a subset of UK Biobank participants. This algorithm may represent a valuable tool for clinicians allowing for fast and potentially less operator-dependent myocardial tissue characterisation. However, validation of more extensive datasets and quality control processes are needed.

Higher consumption of red and processed meat is associated with adverse cardiovascular magnetic resonance morpho-functional phenotypes: A study of 19,408 UK Biobank participants

Funding Acknowledgements

Type of funding sources: Foundation. Main funding source(s): British Heart Foundation Clinical Research Training Fellowship No. FS/17/81/33318 European Union’s Horizon 2020 research and innovation programme under grant agreement No 825903 (euCanSHare project).

Background

Multiple epidemiological studies link greater red and processed meat consumption with poorer cardiovascular outcomes. However, the impact of these exposures on directly measured cardiovascular phenotypes has not been examined in large cohorts. Limited existing studies suggest that the observed associations may be mediated by cardiometabolic diseases and/or novel mechanisms acting via the heart-gut axes. However, few studies systematically examine potential confounding and mediating mechanisms.

Purpose

We assessed, in the UK Biobank, the association between meat intake and cardiovascular structure and function incorporating a comprehensive range of confounders and mediators.

Methods

We studied 19,408 participants with cardiovascular magnetic resonance (CMR) data. We determined average daily red and processed meat intake using food frequency questionnaires. We used oily fish as a comparator linked to favourable cardiac health. We considered conventional CMR measures (ventricular volumes, left ventricular mass, ejection fraction, stroke volume), novel CMR radiomics features (shape, texture), and arterial stiffness metrics (arterial stiffness index, aortic distensibility). Multivariable linear regression models were used to investigate associations between meat/fish intake and cardiovascular phenotypes, adjusting for age, sex, deprivation, educational level, smoking, alcohol intake, and exercise. In separate models, we investigated the mediating role of cardiometabolic morbidities.

Results

Higher intake of red and processed meat was associated with an adverse overall pattern of right and left ventricular remodelling, poorer cardiac function, and higher arterial stiffness. Conversely, higher oily fish intake was associated with a healthy cardiovascular phenotype (better ventricular function, greater arterial compliance). Radiomics analysis showed association of the different dietary habits with unique overall geometry of the ventricles and myocardial texture. These associations were partially mediated by cardiometabolic morbidities.

Conclusions

Higher red and processed meat consumption is associated with adverse cardiovascular phenotypes. These relationships are not fully explained by mediation through cardiometabolic morbidities suggesting importance of alternative disease pathways. Understanding these potential novel disease mechanisms is important for optimising cardiovascular disease prevention strategies.

Application of a machine learning contouring tool for the evaluation of left ventricular strain in clinical practice

Funding Acknowledgements

Type of funding sources: Private grant(s) and/or Sponsorship. Main funding source(s): AK has been funded by the Egyptian cultural centre and educational bureau of the Egyptian embassy in London and the Ministry of higher education in Egypt. SEP acknowledges support from the “SmartHeart” EPSRC programme grant (www.nihr.ac.uk; EP/P001009/1) and the London Medical Imaging and AI Centre for Value-Based Healthcare. This new centre is one of the UK Centres supported by a £50m investment from the Data to Early Diagnosis and Precision Medicine strand of the government’s Industrial Strategy Challenge Fund, managed and delivered by UK Research and Innovation (UKRI). SEP acknowledges support from the CAP-AI programme, London’s first AI enabling programme focused on stimulating growth in the capital’s AI Sector. CAP-AI is led by Capital Enterprise in partnership with Barts Health NHS Trust and Digital Catapult and is funded by the European Regional Development Fund and Barts Charity. SEP also acts as a paid consultant to Circle Cardiovascular Imaging Inc., Calgary, Canada and Servier

onbehalf

Barts Heart Centre, St Bartholomew’s Hospital, Barts Health NHS Trust, West Smithfield, London, UK

Background

Manual contouring of cardiovascular magnetic resonance (CMR) cine images remains common practice and the reference standard for left ventricular (LV) volumes and mass evaluation. However, it is time-consuming and machine learning (ML) may significantly reduce the time required for contouring. Accurate LV contours are the basis for reliable LV strain analysis using tissue tracking.

Purpose

To assess the impact of a ML contouring tool alone versus expert adjusted contours on LV strain.

Methods

We retrospectively selected 402 CMR studies with diagnoses of myocardial infarction (n = 108), myocarditis (n = 130) and healthy controls (n = 164) from the Barts BioResource between January 2015 to June 2018. CMR examinations were obtained using 1.5T and 3T scanners (Siemens Healthineers, Germany). We excluded 32 cases due to phase inconsistency between short (SAX) and long axes (LAX) cine images or suboptimal cine image quality. For the remaining 370 cases, steady state free precession cine images for LAX and SAX were analysed by the ML contouring tool (using CVI42 research prototype software 5.11). Manual expert adjustment for the contours was done for each case if considered suboptimal for strain analysis in the reference end-diastolic phase. Strain results from ML and expert adjusted ML methods were compared for strain agreement. Times taken by these methods were recorded and compared against the time taken for standard manual contouring.

Results

SAX and LAX derived strains by ML and expert adjusted ML methods showed good agreement by Bland-Altman analysis (Figure 1) with excellent coefficient of concordance using Kendall W which is 0.98 for global SAX, radial and circumferential strains (mean difference(MD) = -1.7% (lower and upper limits of agreement (UL,LL) -6.6,3.2), MD = 0.5% (-1.0,2.1)) and is 0.95 for global LAX derived strain (radial and longitudinal, MD = 0.7% (UL,LL -8.7 ,7.4),MD= 0.2% (-1.9,2.5), respectively). Time taken for adjustment of ML contours was significantly shorter than manual contouring (1.35 minutes vs 8.0 minutes, around 590% time saving in ML adjusted method).

Conclusions

ML contouring compared to expert manual adjustment has a clinically reasonable agreement when used for measuring LV strain. Also, using the ML tool with expert adjustment shows significant time saving for analysis and reporting time compared to entirely manual analysis, favouring its application in routine clinical practice.

Abstract Figure.

Assessing automated CMR contouring algorithms using systematic contour quality scoring analysis

Funding Acknowledgements

Type of funding sources: Public grant(s) – National budget only. Main funding source(s): SmartHeart EPSRC programme grant (www.nihr.ac.uk), London Medical Imaging and AI Centre for Value-Based Healthcare

Background

Quality measures for machine learning algorithms include clinical measures such as end-diastolic (ED) and end-systolic (ES) volume, volumetric overlaps such as Dice similarity coefficient and surface distances such as Hausdorff distance. These measures capture differences between manually drawn and automated contours but fail to capture the trust of a clinician to an automatically generated contour.

Purpose

We propose to directly capture clinicians’ trust in a systematic way. We display manual and automated contours sequentially in random order and ask the clinicians to score the contour quality. We then perform statistical analysis for both sources of contours and stratify results based on contour type.

Data

The data selected for this experiment came from the National Health Center Singapore. It constitutes CMR scans from 313 patients with diverse pathologies including: healthy, dilated cardiomyopathy (DCM), hypertension (HTN), hypertrophic cardiomyopathy (HCM), ischemic heart disease (IHD), left ventricular non-compaction (LVNC), and myocarditis. Each study contains a short axis (SAX) stack, with ED and ES phases manually annotated. Automated contours are generated for each SAX image for which manual annotation is available. For this, a machine learning algorithm trained at Circle Cardiovascular Imaging Inc. is applied and the resulting predictions are saved to be displayed in the contour quality scoring (CQS) application.

Methods: The CQS application displays manual and automated contours in a random order and presents the user an option to assign a contour quality score

1: Unacceptable, 2: Bad, 3: Fair, 4: Good. The UK Biobank standard operating procedure is used for assessing the quality of the contoured images. Quality scores are assigned based on how the contour affects clinical outcomes. However, as images are presented independent of spatiotemporal context, contour quality is assessed based on how well the area of the delineated structure is approximated. Consequently, small contours and small deviations are rarely assigned a quality score of less than 2, as they are not clinically relevant. Special attention is given to the RV-endo contours as often, mostly in basal images, two separate contours appear. In such cases, a score of 3 is given if the two disjoint contours sufficiently encompass the underlying anatomy; otherwise they are scored as 2 or 1.

Results

A total of 50991 quality scores (24208 manual and 26783 automated) are generated by five expert raters. The mean score for all manual and automated contours are 3.77 ± 0.48 and 3.77 ± 0.52, respectively. The breakdown of mean quality scores by contour type is included in Fig. 1a while the distribution of quality scores for various raters are shown in Fig. 1b.

Conclusion

We proposed a method of comparing the quality of manual versus automated contouring methods. Results suggest similar statistics in quality scores for both sources of contours.

Abstract Figure 1

Inline perfusion mapping provides insights into the disease mechanism in hypertrophic cardiomyopathy

Objective

In patients with hypertrophic cardiomyopathy (HCM), the role of small vessel disease and myocardial perfusion remains incompletely understood and data on absolute myocardial blood flow (MBF, mL/g/min) are scarce. We measured MBF using cardiovascular magnetic resonance fully quantitative perfusion mapping to determine the relationship between perfusion, hypertrophy and late gadolinium enhancement (LGE) in HCM.

Methods

101 patients with HCM with unobstructed epicardial coronary arteries and 30 controls (with matched cardiovascular risk factors) underwent pixel-wise perfusion mapping during adenosine stress and rest. Stress, rest MBF and the myocardial perfusion reserve (MPR, ratio of stress to rest) were calculated globally and segmentally and then associated with segmental wall thickness and LGE.

Results

In HCM, 79% had a perfusion defect on clinical read. Stress MBF and MPR were reduced compared with controls (mean±SD 1.63±0.60 vs 2.30±0.64 mL/g/min, p<0.0001 and 2.21±0.87 vs 2.90±0.90, p=0.0003, respectively). Globally, stress MBF fell with increasing indexed left ventricle mass (R² for the model 0.186, p=0.036) and segmentally with increasing wall thickness and LGE (both p<0.0001). In 21% of patients with HCM, MBF was lower during stress than rest (MPR <1) in at least one myocardial segment, a phenomenon which was predominantly subendocardial. Apparently normal HCM segments (normal wall thickness, no LGE) had reduced stress MBF and MPR compared with controls (mean±SD 1.88±0.81 mL/g/min vs 2.32±0.78 mL/g/min, p<0.0001).

Conclusions

Microvascular dysfunction is common in HCM and associated with hypertrophy and LGE. Perfusion can fall during vasodilator stress and is abnormal even in apparently normal myocardium suggesting it may be an early disease marker.

P2249Does self-reported pregnancy loss identify women at risk of an adverse cardiovascular phenotype in later life? Insights from UK biobank

Introduction

Cardiovascular disease (CVD) is more common in women who have had pregnancy complications such as spontaneous pregnancy loss. We used cross-sectional data from the UK Biobank Imaging Enhancement Study to determine whether pregnancy loss is associated with cardiac or vascular remodelling in later life, which might contribute to this increased risk.

Methods

Pregnancy history was reported by women participating in UK Biobank between 2006 and 2010 at age 40–69 years using a self-completed touch-screen questionnaire. Self-reported pregnancy loss was related to cardiovascular measures collected in those women who had participated in the Imaging Enhancement Study up to the end of 2015. Cardiac structure and function were assessed by magnetic resonance (CMR) steady-state free precession imaging at 1.5 Tesla. Three long axes cines (horizontal, vertical and LV outflow tract) and a complete short axis stack were acquired, covering both ventricles. Tagging was used to measure myocardial strain in basal, midventricular and apical short axes views. Carotid intima-media thickness (CIMT) measurements were taken for both common carotid arteries using a CardioHealth Station. Statistical associations with CMR and carotid measures were adjusted for age, BMI and other cardiovascular risk factors.

Results

Data were available on 2660 women of whom 113 were excluded because of pre-existing CVD and 8 had no pregnancy information available. Of the remaining 2539, 466 were nulligravid and 2073 had a history of pregnancies, of whom 622 reported at least one pregnancy loss (92% miscarriages and 8% stillbirths) and 1451 reported no pregnancy loss. No significant differences in cardiac or carotid parameters were evident in women who reported pregnancy loss compared to other groups (Table 1).

CMR cardiac geometry & CIMT measurements Variable Pregnancy History Adjusted Means ± SE Effect Size (%) 95% CI P LVEDV (ml) Pregnancy Loss 122.2±1.0 0 – – No Pregnancy 124.1±1.4 1.58 (−0.83, 4.05) 0.20 Pregnancy (no loss) 122.2±0.8 0.2 (−1.42, 1.48) 0.97 LVESV (ml) Pregnancy Loss 47.8±0.6 0 – – No Pregnancy 48.0±0.8 0.45 (−3.19, 4.22) 0.81 Pregnancy (no loss) 47.3±0.5 −1.01 (−3.19, 1.22) 0.37 VEF (%) Pregnancy Loss 60.6±0.3 0 – – No Pregnancy 61.0±0.4 0.42 (−0.50, 1.35) 0.37 Pregnancy (no loss) 61.0±0.2 0.43 (−0.14, 0.99) 0.14 LVM (g) Pregnancy Loss 70.6±0.6 0 – – No Pregnancy 70.5±0.8 −0.15 (−2.68, 2.44) 0.91 Pregnancy (no loss) 70.4±0.5 −0.26 (−1.81, 1.30) 0.74 CIMT (μm) Pregnancy Loss 633.3±6.5 0 – – No Pregnancy 619.3±8.4 −2.22 (−5.04, 0.68) 0.13 Pregnancy (no loss) 627.1±4.9 −0.99 (−2.75, 0.81) 0.28

Conclusion

Women who self-report pregnancy loss do not have significant differences in cardiac or carotid structure in later life to explain past epidemiological findings of increased cardiovascular risk in this population. This may be because this risk operates through other disease mechanisms or that self-report is not a sufficiently reliable way to identify pregnancy loss, and thereby allocate women into risk groups.

Influence of Cardiac CT based disease severity and clinical symptoms on the diagnostic performance of myocardial perfusion

We aimed to identify factors influencing the sensitivity of perfusion imaging after an initial positive coronary computed tomography angiography (CCTA) using invasive coronary angiography (ICA) with conditional fractional flow reserve (FFR) as reference. Secondly we aimed to identify factors associated with revascularisation and to evaluate treatment outcome after ICA. We analysed 292 consecutive patients with suspected significant coronary artery disease (CAD) at CCTA, who underwent perfusion imaging with either cardiac magnetic resonance (CMR) or myocardial perfusion scintigraphy (MPS) followed by ICA with conditional FFR. Stratified analysis and uni- and multiple logistic regression analyses were performed to identify predictors of diagnostic agreement between perfusion scans and ICA and predictors of revascularisation. Myocardial ischemia evaluated with perfusion scans was present in 65/292 (22%) while 117/292 (40%) had obstructive CAD evaluated by ICA. Revascularisation rate was 90/292 (31%). The overall sensitivity for perfusion scans was 39% (30-48), specificity 89% (83-93), PPV 69% (57-80) and NPV 68% (62-74). Stratified analysis showed higher sensitivities in patients with multi-vessel disease at CCTA 49% (37-60) and typical chest pain 50% (37-60). Predictors of revascularisation were multi-vessel disease by CCTA (OR 3.51 [1.91-6.48]) and a positive perfusion scan (OR 4.69 [2.49-8.83]). The sensitivity for perfusion scans after CCTA was highest in patients with typical angina and multiple lesions at CCTA and predicted diagnostic agreement between perfusion scans and ICA. Abnormal perfusion and multi vessel disease at CCTA predicted revascularisation.

Using patient-reported outcome measures for primary percutaneous coronary intervention

Introduction

Routine measurement of the outcome of myocardial infarction is usually limited to immediate morbidity and mortality. Our aim was to determine the response to patient-reported outcome measures (PROMs) 3 months later, identify response bias and explore the feasibility of comparing outcome with their recalled view of their prior health state.

Methods

Patients admitted with ST-segment-elevation myocardial infarction (STEMI) to five percutaneous coronary intervention centres were invited to complete a retrospective questionnaire containing the EQ-5D-3L and short form Seattle Angina Questionnaire (SAQ-7). Response rate for a 3-month mailed follow-up questionnaire and potential response biases were assessed. Patients’ outcomes were compared with their baseline using χ² and paired t-test to assess for differences.

Results

Of 392 patients contacted, 260 (66.3%) responded. Responders were more likely to be older, female, more affluent and have a higher EQ-5D at baseline. Three months after surgery, patients’ SAQ-7 and angina symptom subscale returned to their baseline score. The physical limitation subscale score was worse than at baseline (79.9 vs 73.2, p=0.002), whereas the quality-of-life subscale was better (66.6 vs 73.9; p<0.001). The EQ-5D-3L index score was similar at 3 months to baseline (0.82 vs 0.79). Evidence of bias arising from responders being in better general health at baseline needs further investigation and, if confirmed, needs to be taken into account in interpreting PROMs data.

Conclusion

It is feasible to use PROMs routinely to assess the impact of emergency admissions of patients with STEMI. A larger demonstration project with more sites is needed to confirm these findings.

Prior fragility fracture and risk of incident ischaemic cardiovascular events: results from UK Biobank

In the large UK Biobank population-based cohort, we found that amongst men, but not women, prior fragility fracture was associated with increased risk of admission with ischaemic heart disease.

INTRODUCTION

We aimed to investigate the relationship between prior fracture and risk of incident ischaemic cardiovascular events in a UK population-based cohort.

METHODS

UK Biobank is a large prospective cohort comprising 502,637 men and women aged 40-69 years, with detailed baseline assessment. History of fracture was self-reported, and details of hospital admissions for ischaemic heart disease (IHD) (ICD-10:I20-I25) were obtained through linkage to UK Hospital Episode Statistics. Cox proportional hazards models were used to investigate the prospective relationships between prior fracture and hospital admission for men and women, controlling for age, BMI, smoking, alcohol, educational level, physical activity, systolic blood pressure, calcium and vitamin D use, ankle spacing-width, heel BUA and HRT use (women).

RESULTS

Amongst men, a fragility fracture (hip, spine, wrist or arm fracture resulting from a simple fall) within the previous 5 years was associated with a 35% increased risk of IHD admission (fully adjusted HR 1.35; 95%CI 1.00, 1.82; p = 0.047), with the relationship predominantly driven by wrist fractures. Associations with hospitalisation for angina in men were similar in age-adjusted models [HR1.54; 95%CI: 1.03, 2.30), p = 0.037], but did not remain statistical significant after full adjustment [HR 1.64; 95%CI: 0.88, 3.07); p = 0.121]. HRs for admission with angina were lower in women, and neither age- nor fully adjusted relationships attained statistical significance.

CONCLUSIONS

Prior fragility fracture is an independent risk factor for incident ischaemic cardiovascular events in men. Further work may clarify whether this association is causal or represents shared risk factors, but these findings are likely to be of value in risk assessment of both osteoporosis and cardiovascular disease.

Distinct Stages of Moment-to-Moment Processing in the Cinguloopercular and Frontoparietal Networks

Control of goal-directed tasks is putatively carried out via the cinguloopercular (CO) and frontoparietal (FP) systems. However, it remains unclear whether these systems show dissociable moment-to-moment processing during distinct stages of a trial. Here, we characterize dynamics in the CO and FP networks in a meta-analysis of 5 decision-making tasks using fMRI, with a specialized "slow reveal" paradigm which allows us to measure the temporal characteristics of trial responses. We find that activations in left FP, right FP, and CO systems form separate clusters, pointing to distinct roles in decision-making. Left FP shows early "accumulator-like" responses, suggesting a role in pre-decision processing. CO has a late onset and transient response linked to the decision event, suggesting a role in performance reporting. The majority of right FP regions show late onsets with prolonged responses, suggesting a role in post-recognition processing. These findings expand upon past models, arguing that the CO and FP systems relate to distinct stages of processing within a trial. Furthermore, the findings provide evidence for a heterogeneous profile in the FP network, with left and right FP taking on specialized roles. This evidence informs our understanding of how distinct control networks may coordinate moment-to-moment components of complex actions.

Influence of image slice thickness on rectal dose-response relationships following radiotherapy of prostate cancer

When pooling retrospective data from different cohorts, slice thicknesses of acquired computed tomography (CT) images used for treatment planning may vary between cohorts. It is, however, not known if varying slice thickness influences derived dose-response relationships. We investigated this for rectal bleeding using dose-volume histograms (DVHs) of the rectum and rectal wall for dose distributions superimposed on images with varying CT slice thicknesses. We used dose and endpoint data from two prostate cancer cohorts treated with three-dimensional conformal radiotherapy to either 74 Gy (N = 159) or 78 Gy (N = 159) at 2 Gy per fraction. The rectum was defined as the whole organ with content, and the morbidity cut-off was Grade ≥2 late rectal bleeding. Rectal walls were defined as 3 mm inner margins added to the rectum. DVHs for simulated slice thicknesses from 3 to 13 mm were compared to DVHs for the originally acquired slice thicknesses at 3 and 5 mm. Volumes, mean, and maximum doses were assessed from the DVHs, and generalized equivalent uniform dose (gEUD) values were calculated. For each organ and each of the simulated slice thicknesses, we performed predictive modeling of late rectal bleeding using the Lyman-Kutcher-Burman (LKB) model. For the most coarse slice thickness, rectal volumes increased (≤18%), whereas maximum and mean doses decreased (≤0.8 and ≤4.2 Gy, respectively). For all a values, the gEUD for the simulated DVHs were ≤1.9 Gy different than the gEUD for the original DVHs. The best-fitting LKB model parameter values with 95% CIs were consistent between all DVHs. In conclusion, we found that the investigated slice thickness variations had minimal impact on rectal dose-response estimations. From the perspective of predictive modeling, our results suggest that variations within 10 mm in slice thickness between cohorts are unlikely to be a limiting factor when pooling multi-institutional rectal dose data that include slice thickness variations within this range.