Impact of image quality on reliability of the measurements of left ventricular systolic function and global longitudinal strain in 2D echocardiography

A multi-task deep learning approach for real-time view classification and quality assessment of echocardiographic images

High-quality standard views in two-dimensional echocardiography are essential for accurate cardiovascular disease diagnosis and treatment decisions. However, the quality of echocardiographic images is highly dependent on the practitioner's experience. Ensuring timely quality control of echocardiographic images in the clinical setting remains a significant challenge. In this study, we aimed to propose new quality assessment criteria and develop a multi-task deep learning model for real-time multi-view classification and image quality assessment (six standard views and "others"). A total of 170,311 echocardiographic images collected between 2015 and 2022 were utilized to develop and evaluate the model. On the test set, the model achieved an overall classification accuracy of 97.8% (95%CI 97.7-98.0) and a mean absolute error of 6.54 (95%CI 6.43-6.66). A single-frame inference time of 2.8 ms was achieved, meeting real-time requirements. We also analyzed pre-stored images from three distinct groups of echocardiographers (junior, senior, and expert) to evaluate the clinical feasibility of the model. Our multi-task model can provide objective, reproducible, and clinically significant view quality assessment results for echocardiographic images, potentially optimizing the clinical image acquisition process and improving AI-assisted diagnosis accuracy.

Prognostic value of fully-automated left atrial strain in patients with asymptomatic chronic severe aortic regurgitation

BACKGROUND

To examine whether left atrial (LA) strain was associated with adverse outcomes in asymptomatic chronic aortic regurgitation (AR).

METHODS

Asymptomatic patients with ≥moderate-severe AR were retrospectively identified from 2008 through 2022 from a university hospital. Apical 4-chamber left ventricular longitudinal strain (A4C-LVLS), LA reservoir (LASr), conduit (LAScd), and contractile strain (LASct) were measured using fully-automated software. Primary endpoint was all-cause death (ACD); secondary endpoints were heart failure (HF) development or aortic valve surgery (AVS).

RESULTS

Of 352 patients (59 ± 17 years; 19 % female), the mean LV ejection fraction (LVEF) was 60 ± 8 %. The median follow-up during medical surveillance was 4.7 (interquartile range: 1.8-9.0) years; during which 68 patients died. Multivariable analysis adjusted for covariates showed that larger maximal LA volume index (iLAVmax), lower LASr and LASct were independently associated with ACD (all P ≤ 0.047); A4C-LVLS and LAScd were not (P ≥ 0.15). Besides, iLAVmax, LASr, and LASct provided incremental prognostic value over A4C-LVLS in terms of ACD (all P ≤ 0.048). HF symptoms occurred in 126 patients at a median of 2 years. Multivariable determinants for HF development included larger minimal LAV index, lower LASr and LASct (all P ≤ 0.03). Adjusted spline curves showed LASr <38-40 % and LASct <20-24 % were associated with increased risks of ACD and HF development, respectively. Using abovementioned LASr and LASct cutoffs, adjusted Kaplan-Meier curves risk-stratified patients for ACD successfully (P ≤ 0.02). Lower LASr was also independently associated with AVS (Hazard ratio per 1 % increase: 0.98)(P = 0.02).

CONCLUSIONS

In patients with asymptomatic AR, fully-automated LASr and LASct were robust markers for outcome determination; these markers may identify those who need timely surgical referral.

Machine Learning and Deep Learning Methods for Fast and Accurate Assessment of Transthoracic Echocardiogram Image Quality

High-quality echocardiogram images are the cornerstone of accurate and reliable measurements of the heart. Therefore, this study aimed to develop, validate and compare machine learning and deep learning algorithms for accurate and automated assessment of transthoracic echocardiogram image quality. In total, 4090 single-frame two-dimensional transthoracic echocardiogram images were used from apical 4-chamber, apical 2-chamber and parasternal long-axis views sampled from 3530 adult patients. The data were extracted from CAMUS and Unity Imaging open-source datasets. For every raw image, additional grayscale block histograms were developed. For block histogram datasets, six classic machine learning algorithms were tested. Moreover, convolutional neural networks based on the pre-trained EfficientNetB4 architecture were developed for raw image datasets. Classic machine learning algorithms predicted image quality with 0.74 to 0.92 accuracy (AUC 0.81 to 0.96), whereas convolutional neural networks achieved between 0.74 and 0.89 prediction accuracy (AUC 0.79 to 0.95). Both approaches are accurate methods of echocardiogram image quality assessment. Moreover, this study is a proof of concept of a novel method of training classic machine learning algorithms on block histograms calculated from raw images. Automated echocardiogram image quality assessment methods may provide additional relevant information to the echocardiographer in daily clinical practice and improve reliability in clinical decision making.

Multi-Ethnic Study of Atherosclerosis Early Heart Failure Study: Rationale, Design, and Baseline Characteristics

BACKGROUND

Current prevalence estimates of heart failure (HF) are primarily based on self-report or HF hospitalizations. There is an unmet need to define the prevalence and pathogenesis of early symptomatic HF, which may be undiagnosed and precedes HF hospitalization.

METHODS

The MESA (Multi-Ethnic Study of Atherosclerosis) Early HF study was conducted during MESA exam 6 to determine the prevalence of early HF and investigate the transition from risk factors to early HF in a diverse population-based cohort of older adults. Between 2016 and 2018, 3285 MESA participants from 6 field centers underwent comprehensive speckle-tracking echocardiography with passive leg raise maneuver, Kansas City Cardiomyopathy Questionnaire, 6-minute walk test, arterial stiffness assessment, and proteomics (including NT-proBNP [N-terminal pro-B-type natriuretic peptide]).

RESULTS

Median age was 73 (25th-75th percentile 67-81) years, 53.2% were female, 25.6% were Black, 12.8% were Chinese, and 40.0% were White. The prevalence of HF risk factors was high: hypertension, 61.9%; former or current smoking, 53.7%; obesity 34.8%; diabetes; 24.7%; and chronic kidney disease; 22%. Overt cardiovascular disease, which ranged from 2.1% (HF) to 13.6% (atrial fibrillation), was less common. Of the 3285 participants, 96% underwent proteomics, 94% Kansas City Cardiomyopathy Questionnaire, 93% speckle-tracking echocardiography with passive leg raise, 82% arterial stiffness exam, and 77% 6-minute walk test. Feasibility of resting speckle-tracking echocardiography (87%-99% across cardiac chambers) and passive leg raise Doppler/speckle-tracking echocardiography (>84%) measurements was high. A total of 120 unique echocardiographic indices were measured.

CONCLUSIONS

The MESA Early HF study is a key resource for cardiovascular researchers who are interested in improving the epidemiological and phenotypic characterization of early HF.

REGISTRATION

URL: https://www.clinicaltrials.gov; Unique identifier: NCT00005487.

Influence of ultrasound transmit frequency on measurement of global longitudinal strain on 2D speckle tracking echocardiography

The reproducibility of longitudinal strain measured by 2D speckle tracking echocardiography (2DSTE) may be affected by ultrasound settings. This study investigated the effect of transmit ultrasound frequency on global longitudinal strain (GLS) by 2DSTE. Apical, 2- and 4-chamber, and long-axis views were obtained in consecutive 162 patients using Philips ultrasound devices. Three different frequency presets were used sequentially: high resolution (HRES, 1.9 to 2.1 MHz), general (HGEN, 1.6 to 1.8 MHz), and penetration mode (HPEN, 1.3 to 1.6 MHz). GLS values were determined for each preset using the Philips Q-station software, resulting in GLS-HRES, GLS-HGEN, and GLS-HPEN. Among the 151 patients with successfully measured GLS, a significant difference in GLS was observed among the three presets (p < 0.0001). GLS-HRES (- 17.9 ± 4.4%) showed a slightly smaller magnitude compared to GLS-HGEN (- 18.8 ± 4.5%, p < 0.0001) and GLS-HPEN (- 18.8 ± 4.5%, p < 0.0001), with absolute differences of 1.1 ± 1.0% and 1.1 ± 1.2%, respectively. This variation in GLS with frequency was evident in patients with both optimal (n = 104) and suboptimal (n = 47) image quality and remained consistent regardless of ultrasound devices, ischemic etiology, or ejection fraction. In conclusion, ultrasound frequency had only a modest effect on GLS measurements. GLS may be reliably assessed in most cases regardless of the ultrasound frequency used.

Stress-Mediated Abnormalities in Regional Myocardial Wall Motion in Young Women with a History of Psychological Trauma

BACKGROUND

Psychosocial stress has been associated with the development and progression of atherosclerotic cardiovascular disease (CVD). Previously, we reported subtle differences in global longitudinal strain in somatically healthy women with a psychiatric diagnosis of borderline personality disorder (BPD). This study aimed to investigate the impact of BPD on segmental myocardial wall motion using speckle tracking echocardiography (STE) analysis.

METHODS

A total of 100 women aged between 18 and 38 years were included in this study. Fifty patients meeting the diagnostic criteria for BPD were recruited from the Department of Psychiatry (LWL-University Hospital Bochum) and compared with fifty age-matched healthy control subjects without previous cardiac disease. Laboratory tests and STE were performed with segmental wall motion analysis.

RESULTS

The BPD group had a higher prevalence of risk factors for CVD, with smoking and obesity being predominant, when compared with the control group. Other cardiovascular parameters such as blood pressure, glucose, and cholesterol levels were also elevated, even though not to pathological values. Moreover, in the STE analysis, the BPD group consistently exhibited decreased deformation in nine myocardial wall regions compared with the control group, along with a shift toward higher values in the distribution of peak pathological segments. Additionally, significantly higher values of free thyroxine concentration and thyroid's secretory capacity were observed in the BPD group, despite falling within the (high-) normal range.

CONCLUSIONS

BPD is associated with chronic stress, classical risk factors, and myocardial wall motion abnormalities. Further exploration is warranted to investigate the relationship between high-normal thyroid metabolism, these risk factors, and myocardial function in BPD patients. Long-term follow-up studies would be valuable in confirming the potential for predicting adverse events.

Embracing AI: The Imperative Tool for Echo Labs to Stay Ahead of the Curve

Advancements in artificial intelligence (AI) have rapidly transformed various sectors, and the field of echocardiography is no exception. AI-driven technologies hold immense potential to revolutionize echo labs' diagnostic capabilities and improve patient care. This paper explores the importance for echo labs to embrace AI and stay ahead of the curve in harnessing its power. Our manuscript provides an overview of the growing impact of AI on medical imaging, specifically echocardiography. It highlights how AI-driven algorithms can enhance image quality, automate measurements, and accurately diagnose cardiovascular diseases. Additionally, we emphasize the importance of training echo lab professionals in AI implementation to optimize its integration into routine clinical practice. By embracing AI, echo labs can overcome challenges such as workload burden and diagnostic accuracy variability, improving efficiency and patient outcomes. This paper highlights the need for collaboration between echocardiography laboratory experts, AI researchers, and industry stakeholders to drive innovation and establish standardized protocols for implementing AI in echocardiography. In conclusion, this article emphasizes the importance of AI adoption in echocardiography labs, urging practitioners to proactively integrate AI technologies into their workflow and take advantage of their present opportunities. Embracing AI is not just a choice but an imperative for echo labs to maintain their leadership and excel in delivering state-of-the-art cardiac care in the era of advanced medical technologies.

Automated Detection of Apical Foreshortening in Echocardiography Using Statistical Shape Modelling

OBJECTIVE

Automated detection of foreshortening, a common challenge in routine 2-D echocardiography, has the potential to improve quality of acquisitions and reduce the variability of left ventricular measurements. Acquiring and labelling the required training data is challenging due to the time-intensive and highly subjective nature of foreshortened apical views. We aimed to develop an automatic pipeline for the detection of foreshortening. To this end, we propose a method to generate synthetic apical-four-chamber (A4C) views with matching ground truth foreshortening labels.

METHODS

A statistical shape model of the four chambers of the heart was used to synthesise idealised A4C views with varying degrees of foreshortening. Contours of the left ventricular endocardium were segmented in the images, and a partial least squares (PLS) model was trained to learn the morphological traits of foreshortening. The predictive capability of the learned synthetic features was evaluated on an independent set of manually labelled and automatically curated real echocardiographic A4C images.

RESULTS

Acceptable classification accuracy for identification of foreshortened views in the testing set was achieved using logistic regression based on 11 PLS shape modes, with a sensitivity, specificity and area under the receiver operating characteristic curve of 0.84, 0.82 and 0.84, respectively. Both synthetic and real cohorts showed interpretable traits of foreshortening within the first two PLS shape modes, reflected as a shortening in the long-axis length and apical rounding.

CONCLUSION

A contour shape model trained only on synthesized A4C views allowed accurate prediction of foreshortening in real echocardiographic images.

Machine learning derived echocardiographic image quality in patients with left ventricular systolic dysfunction: insights on the echo views of greatest image quality

We sought to determine the cardiac ultrasound view of greatest quality using a machine learning (ML) approach on a cohort of transthoracic echocardiograms (TTE) with abnormal left ventricular (LV) systolic function. We utilize an ML model to determine the TTE view of highest quality when scanned by sonographers. A random sample of TTEs with reported LV dysfunction from 09/25/2017-01/15/2019 were downloaded from the regional database. Component video files were analyzed using ML models that jointly classified view and image quality. The model consisted of convolutional layers for extracting spatial features and Long Short-term Memory units to temporally aggregate the frame-wise spatial embeddings. We report the view-specific quality scores for each TTE. Pair-wise comparisons amongst views were performed with Wilcoxon signed-rank test. Of 1,145 TTEs analyzed by the ML model, 74.5% were from males and mean LV ejection fraction was 43.1 ± 9.9%. Maximum quality score was best for the apical 4 chamber (AP4) view (70.6 ± 13.9%, p<0.001 compared to all other views) and worst for the apical 2 chamber (AP2) view (60.4 ± 15.4%, p<0.001 for all views except parasternal short-axis view at mitral/papillary muscle level, PSAX M/PM). In TTEs scanned by professional sonographers, the view with greatest ML-derived quality was the AP4 view.

Preliminary Feasibility of Stress Myocardial Elastography for the Detection of Coronary Artery Disease

Myocardial elastography (ME) is a cardiac strain imaging technique that has been found capable of detecting a decrease in radial strain caused by ischemia or infarction in patients with coronary artery disease (CAD) as well as in a canine model. Prior studies have focused on rest imaging, but stress testing can reveal functional deficits caused by stenoses that are asymptomatic at rest. Therefore, it has been proposed that stress ME (S-ME) improves the detection of CAD. A novel strain difference (Δε) metric is presented and investigated in a canine model of induced ischemia, as well as in a study in human patients with CAD validated by myocardial perfusion imaging. In the canine model study, flow-limiting stenosis was induced by partial ligation in n = 2 canines, and stenosis was found to consistently reduce Δε in the affected myocardial regions compared with baseline, as well as compared to myocardial regions that are remote to the induced stenosis. In the clinical study, the median Δε was significantly lower (p < 0.05) in infarcted myocardial regions (-6.29%) than in those with normal perfusion (4.62%), with Δε in ischemic regions falling in between (-2.91%). The same trend was observed when considering radial strain during stress and, to a lesser degree, at rest alone. The results indicate that S-ME may be more sensitive to mild cases of CAD that are functionally asymptomatic at rest.

Comparison of the Quality of Echocardiography Imaging Between the Left Lateral Decubitus and Supine Positions

Introduction It is commonly taught that positioning the patient in the left lateral decubitus (LLD) position will improve transthoracic echocardiography (TTE) image quality. Despite this, no previous studies have been performed that study this practice. Our goal was to quantify the difference in image quality of TTE views between the supine and LLD positions.  Methods This was a prospective study in a single academic Emergency Department (ED) of a convenience sample of 30 patients. Three separate ED physicians performed TTE views in both the supine and LLD position on each patient. The order of position was randomized. Images were then reviewed on a previously validated TTE image quality scale by two blinded ED physicians with specialized training in ultrasound. The scale used a 0 to 5 (highest quality) metric for quality assessment. Interpretability of right ventricular and left ventricular function was also assessed. Results The mean image quality for the supine position was 2.85 (standard deviation {SD} 1.1) and 3.05 (SD 1.2) for the LLD position (p=0.044). In the subset of parasternal and apical windows, the mean quality for the supine position was 2.87 (SD 1.1) and 3.23 (SD 1.1) for the LLD position (p=0.003). The number of studies in which right ventricular function was interpretable was significantly higher in the LLD position (62% versus 42%, p=0.044). Conclusions There was a statistically significant increase in image quality when TTE was performed in the LLD position as compared to supine. This was especially pronounced in the apical four and parasternal windows.

Fully Automated Mouse Echocardiography Analysis Using Deep Convolutional Neural Networks

Echocardiography (echo) is a translationally relevant ultrasound imaging modality widely used to assess cardiac structure and function in preclinical models of heart failure (HF) during research and drug development. Though echo is a very valuable tool, the image analysis is a time consuming, resource demanding process, and is susceptible to inter-reader variability. Recent advancements in deep learning have enabled researchers to automate image processing and reduce analysis time and inter-reader variability in the field of medical imaging. In the present study, we developed a fully automated tool - Mouse Echo Neural Net (MENN) - for the analysis of both long axis brightness (B)-mode and short axis motion (M)-mode images of the left ventricle. MENN is a series of fully convolutional neural networks that were trained and validated using manually segmented B-mode and M-mode echo images of the left ventricle. The segmented images were then used to compute cardiac structural and functional metrics. The performance of MENN was further validated in two preclinical models of HF. MENN achieved excellent correlations (Pearson's r = 0.85 to 0.99) and good to excellent agreement between automated and manual analyses. Further inter-reader variability analysis showed that MENN has better agreements with an expert analyst than both a trained analyst and a novice. Notably, the use of MENN reduced manual analysis time by >92%. In conclusion, we developed an automated echocardiography analysis tool that allows for fast and accurate analysis of B-mode and M-mode mouse echo data and mitigates the issue of inter-reader variability in manual analysis.

Phenotyping the hypertensive heart

Arterial hypertension remains the most frequent cardiovascular (CV) risk factor, and is responsible for a huge global burden of disease. Echocardiography is the first-line imaging method for the evaluation of cardiac damage in hypertensive patients and novel techniques, such as 2D and D speckle tracking and myocardial work, provide insight in subclinical left ventricular (LV) impairment that would not be possible to detect with conventional echocardiography. The structural, functional, and mechanical cardiac remodelling that are detected with imaging are intermediate stages in the genesis of CV events, and initiation or intensification of antihypertensive therapy in response to these findings may prevent or delay progressive remodelling and CV events. However, LV remodelling—especially LV hypertrophy—is not specific to hypertensive heart disease (HHD) and there are circumstances when other causes of hypertrophy such as athlete heart, aortic stenosis, or different cardiomyopathies need exclusion. Tissue characterization obtained by LV strain, cardiac magnetic resonance, or computed tomography might significantly help in the distinction of different LV phenotypes, as well as being sensitive to subclinical disease. Selective use of multimodality imaging may therefore improve the detection of HHD and guide treatment to avoid disease progression. The current review summarizes the advanced imaging tests that provide morphological and functional data about the hypertensive cardiac injury.

Echocardiography image quality of global longitudinal strain in cardio-oncology: a prospective real-world investigation

BACKGROUND

Left-ventricular (LV) global longitudinal strain (GLS) has been reported to be a robust and sensitive marker of chemotherapy-induced cardiac damage. Image quality is paramount for accurate GLS measurements. In real-world cardio-oncology settings, the incidence of suboptimal echocardiography quality and its significance in clinical decision-making have not been well investigated. This prospective study examined the incidence and impact of suboptimal echocardiographic image quality on detecting subtle myocardial damage by chemotherapy.

METHODS

Seventy-seven consecutive patients with breast cancer (age, 52 ± 12 years, 76 women, 33 with left-sided breast cancer) were included in this study. Echocardiography was performed at 3-month intervals 1 year before and after chemotherapy initiation. We classified the image quality of each echocardiographic acquisition into three groups: optimal, suboptimal, or inadequate for speckle tracking.

RESULTS

Among the 376 examinations obtained during the cardiac monitoring, the image quality in 194 (52%) was optimal, suboptimal in 159 (42%), and inadequate in 23 (6%). The interobserver reproducibility was 0.91 in the optimal and 0.21 in the suboptimal group. In contrast, the optimal group showed progressive impairment in both GLS (p = 0.001) and LV ejection fraction (LVEF) (p < 0.001) during follow-up, and the suboptimal group showed a progressive decrease in LVEF (p = 0.006), but not in GLS (p = 0.13). Left-sided mammotomy and/or reconstruction surgery and high body mass index were significant determinants of suboptimal image quality.

CONCLUSIONS

Even in cases of minor image quality impairment, the physician should assess GLS carefully to avoid errors in crucial clinical decision-making.

Artificial Intelligence in Echocardiography

Artificial intelligence in diagnostic cardiac-imaging platforms is advancing rapidly. In particular, artificial intelligence algorithms have increased the efficiency and accuracy of echocardiographic cardiovascular imaging, resulting in more complex echocardiographic imaging techniques and expanded use among noncardiologists. Here, we provide an overview of real-world applications of artificial intelligence in echocardiography including automatic high-quality computer-optimized image acquisition sequences, automated measurements, and algorithms for the rapid and accurate interpretation of cardiac physiology. These advances will not replace physicians but will improve their productivity, workflow, and diagnostic performance.

Full cardiac cycle asynchronous temporal compounding of 3D echocardiography images

It is important to improve echocardiography image quality, because the accuracy of echocardiographic assessment and diagnosis relies on image quality. Previous work on 2D temporal image compounding for image frames with matching cardiac phases (synchronous), and for temporally neighbouring image frames (asynchronous) over small ranges of time frames showed good improvement to image quality. Here, we extend this by performing asynchronous temporal compounding to echocardiographic images in 3D, involving all frames within a cardiac cycle, via a robust 3D cardiac motion estimation algorithm to describe the large image deformations. After compounding, the images can be reanimated via the motion model. Various methods of fusing image frames together are tested, including mean, max, and wavelet methods, and outlier rejection algorithms. The compounding algorithm is applied on 3D human adult, porcine adolescent, and human fetal echocardiography images. Results show significant improvements to contrast-to-noise ratio (CNR) and boundary clarity, and significantly decreased variability in manual quantification of cardiac chamber volumes after compounding. Interestingly, compounding can extend the field of view of the echo images, by reconstructing cardiac structures that momentarily exceeded the field of view, using the motion estimation algorithm to calculate their locations outside the field of view during these time periods. Although all compounding methods provide general improvements, the mean method led to blurred boundaries, while the max methods led to high variability of CNR. Outlier rejection algorithms were found to be useful in addressing these weaknesses.

Artificial intelligence and cardiovascular imaging: A win-win combination

Rapid development of artificial intelligence (AI) is gaining grounds in medicine. Its huge impact and inevitable necessity are also reflected in cardiovascular imaging. Although AI would probably never replace doctors, it can significantly support and improve their productivity and diagnostic performance. Many algorithms have already proven useful at all stages of the cardiac imaging chain. Their crucial practical applications include classification, automatic quantification, notification, diagnosis, and risk prediction. Consequently, more reproducible and repeatable studies are obtained, and personalized reports may be available to any patient. Utilization of AI also increases patient safety and decreases healthcare costs. Furthermore, AI is particularly useful for beginners in the field of cardiac imaging as it provides anatomic guidance and interpretation of complex imaging results. In contrast, lack of interpretability and explainability in AI carries a risk of harmful recommendations. This review was aimed at summarizing AI principles, essential execution requirements, and challenges as well as its recent applications in cardiovascular imaging.

Deep learning-based automated left ventricular ejection fraction assessment using 2-D echocardiography

Deep learning (DL) has been applied for automatic left ventricle (LV) ejection fraction (EF) measurement, but the diagnostic performance was rarely evaluated for various phenotypes of heart disease. This study aims to evaluate a new DL algorithm for automated LVEF measurement using two-dimensional echocardiography (2DE) images collected from three centers. The impact of three ultrasound machines and three phenotypes of heart diseases on the automatic LVEF measurement was evaluated. Using 36890 frames of 2DE from 340 patients, we developed a DL algorithm based on U-Net (DPS-Net) and the biplane Simpson's method was applied for LVEF calculation. Results showed a high performance in LV segmentation and LVEF measurement across phenotypes and echo systems by using DPS-Net. Good performance was obtained for LV segmentation when DPS-Net was tested on the CAMUS data set (Dice coefficient of 0.932 and 0.928 for ED and ES). Better performance of LV segmentation in study-wise evaluation was observed by comparing the DPS-Net v2 to the EchoNet-dynamic algorithm (P = 0.008). DPS-Net was associated with high correlations and good agreements for the LVEF measurement. High diagnostic performance was obtained that the area under receiver operator characteristic curve was 0.974, 0.948, 0.968, and 0.972 for normal hearts and disease phenotypes including atrial fibrillation, hypertrophic cardiomyopathy, dilated cardiomyopathy, respectively. High performance was obtained by using DPS-Net in LV detection and LVEF measurement for heart failure with several phenotypes. High performance was observed in a large-scale dataset, suggesting that the DPS-Net was highly adaptive across different echocardiographic systems.NEW & NOTEWORTHY A new strategy of feature extraction and fusion could enhance the accuracy of automatic LVEF assessment based on multiview 2-D echocardiographic sequences. High diagnostic performance for the determination of heart failure was obtained by using DPS-Net in cases with different phenotypes of heart diseases. High performance for left ventricle segmentation was obtained by using DPS-Net, suggesting the potential for a wider range of application in the interpretation of 2DE images.

Correlation of echocardiographic parameters in prone and supine positions in normal adults using a novel approach

BACKGROUND

Transthoracic echocardiography (TTE) in prone position is challenging. Innovative use of transesophageal echocardiography (TEE) probe to perform TTE for such patients has been described; but reproducibility and correlation of the TTE measurements by this technique with those obtained by the standard supine TTE study are still unknown.

METHODS

We enrolled 30 non-COVID-19 individuals, with a mean (SD) age 35 (10.9) years and 11 females, to study the agreement between the transthoracic measurements of the left ventricular (LV), left atrial (LA), aortic dimensions, and ejection fraction (EF) obtained in prone position using an external TEE probe versus the standard supine position using the conventional TTE probe.

RESULTS

There were no significant differences between LV end-diastolic and end-systolic diameters, septal wall thickness, posterior wall thickness, and aortic root dimensions in the prone versus the supine positions, while the mean EF (60.3% vs 63.1%, P = .014) and mean LA dimensions (1.8 vs 1.9 cm/m² , P < .001) were significantly lower in the prone position. The mean time of scans was significantly longer in the prone as compared to the supine position (12.5 vs 4.5 minutes, P < .001). All supine studies had good quality while in the prone position four studies were of poor quality, and one was nondiagnostic.

CONCLUSIONS

Assessment of cardiac dimensions and systolic function in the prone position using transthoracic TEE probe was feasible. LV and aortic dimensions agreed well with the standard TTE in supine position; however, LA dimensions and EF were lower in the prone position.

Quantification of left atrial contractile function using two-dimensional speckle tracking echocardiography in horses after conversion of atrial fibrillation to sinus rhythm

INTRODUCTION/OBJECTIVES

In horses, persistent atrial dysfunction is a prognostic indicator of recurrence of atrial fibrillation (AF). The objective of this study was to quantify left atrial (LA) contractile function in Warmblood horses after successful cardioversion of AF to sinus rhythm, using standard two-dimensional echocardiography (2DE) and two-dimensional speckle tracking (2DST) analyses, and to provide proof-of-concept for use of 2DST to detect LA contractile dysfunction in horses.

ANIMALS, MATERIALS, AND METHODS

Nineteen Warmblood horses with AF and successful transvenous electrical cardioversion (TVEC) or medical cardioversion using quinidine sulfate (QS) were included. Echocardiography was performed at 24 hr, 72 hr, and 1-6 months after cardioversion. Conventional 2DE-derived indices of LA size and function and global longitudinal strain (S_L) and longitudinal strain rate (SR_L) during LA contractile function were measured to detect differences in LA size and function over time and between treatment modalities. The association between 2DE-derived indices and global S_L and SR_L, respectively, and the agreement of the variables to detect LA dysfunction were assessed.

RESULTS

Global S_L and SR_L, as well as active LA fractional area change (FAC) identified cases with LA dysfunction after TVEC and QS cardioversion and proved useful to demonstrate LA functional recovery over time. Agreement between active LA FAC and S_L and SR_L, respectively, to diagnose atrial dysfunction was substantial to fair at 24 hr after cardioversion and decreased with time.

CONCLUSION

This study shows that 2DST is useful to quantify LA contractile dysfunction in horses after cardioversion of AF.

HARP-I: A Harmonic Phase Interpolation Method for the Estimation of Motion From Tagged MR Images

We proposed a novel method called HARP-I, which enhances the estimation of motion from tagged Magnetic Resonance Imaging (MRI). The harmonic phase of the images is unwrapped and treated as noisy measurements of reference coordinates on a deformed domain, obtaining motion with high accuracy using Radial Basis Functions interpolations. Results were compared against Shortest Path HARP Refinement (SP-HR) and Sine-wave Modeling (SinMod), two harmonic image-based techniques for motion estimation from tagged images. HARP-I showed a favorable similarity with both methods under noise-free conditions, whereas a more robust performance was found in the presence of noise. Cardiac strain was better estimated using HARP-I at almost any motion level, giving strain maps with less artifacts. Additionally, HARP-I showed better temporal consistency as a new method was developed to fix phase jumps between frames. In conclusion, HARP-I showed to be a robust method for the estimation of motion and strain under ideal and non-ideal conditions.

Association of Echocardiographic Left Ventricular End-Systolic Volume and Volume-Derived Ejection Fraction With Outcome in Asymptomatic Chronic Aortic Regurgitation

Importance

Volumetric measurements by transthoracic echocardiogram may better reflect left ventricular (LV) remodeling than conventional linear LV dimensions. However, the association of LV volumes with mortality in patients with chronic hemodynamically significant aortic regurgitation (AR) is unknown.

Objective

To assess whether LV volumes and volume-derived LV ejection fraction (Vol-LVEF) are determinants of mortality in AR.

Design, Setting, and Participants

This cohort study included consecutive asymptomatic patients with chronic moderately severe to severe AR from a tertiary referral center (January 2004 through April 2019).

Exposures

Clinical and echocardiographic data were analyzed retrospectively. Aortic regurgitation severity was graded by comprehensive integrated approach. De novo disk-summation method was used to derive LV volumes and Vol-LVEF.

Main Outcome and Measures

Associations between all-cause mortality under medical surveillance and the following LV indexes: linear LV end-systolic dimension index (LVESDi), linear LVEF, LV end-systolic volume index (LVESVi), and Vol-LVEF.

Results

Of 492 asymptomatic patients (mean [SD] age, 60 [17] years; 425 men [86%]), ischemic heart disease prevalence was low (41 [9%]), and 453 (92.1%) had preserved linear LVEF (≥50%) with mean (SD) LVESVi of 41 (15) mL/m2. At a median (interquartile range) of 5.4 (2.5-10.1) years, 66 patients (13.4%) died under medical surveillance; overall survival was not different than the age- and sex-matched general population (P = .55). Separate multivariate models, adjusted for age, sex, Charlson Comorbidity Index, and AR severity, demonstrated that in addition to linear LVEF and LVESDi, LVESVi and Vol-LVEF were independently associated with mortality under surveillance (all P < .046) with similar C statistics (range, 0.83-0.84). Spline curves showed that continuous risks of death started to rise for both linear LVEF and Vol-LVEF less than 60%, LVESVi more than 40 to 45 mL/m2, and LVESDi above 21 to 22 mm/m2. As dichotomized variables, patients with LVESVi more than 45 mL/m2 exhibited increased relative death risk (hazard ratio, 1.93; 95% CI, 1.10-3.38; P = .02) while LVESDi more than 20 mm/m2 did not (P = .32). LVESVi more than 45 mL/m2 showed a decreased survival trend compared with expected population survival.

Conclusions and Relevance

In this large asymptomatic cohort of patients with hemodynamically significant AR, LVESVi and Vol-LVEF worked equally as well as LVESDi and linear LVEF in risk discriminating patients with excess mortality. A LVESVi threshold of 45 mL/m2 or greater was significantly associated with an increased mortality risk.

Inter-vendor variability and reproducibility of subcostal left ventricular longitudinal strain

Left ventricular longitudinal strain (LVLS) performed using subcostal windows is a novel alternative for patients who require strain imaging but have poor apical windows. We investigated the reproducibility and inter-vendor variability of subcostal LVLS. One hundred and twenty-four echocardiographic studies were analysed from 73 women with early stage HER2-positive breast cancer. Speckle tracking strain was performed offline using EchoPAC and TomTec on subcostal 4-chamber and 3-chamber views to obtain subcostal 4-chamber (SC4_LS) and 3-chamber (SC3_LS) LVLS which was then averaged (SCav_LS). Reproducibility of subcostal single chamber and averaged LVLS were assessed. Measurements between platforms were compared. Strain was reported in absolute magnitude. EchoPAC measurements of SC3_LS (20.5 ± 2.4% vs. 21.2 ± 2.5%, p = 0.002) and SCav_LS (20.9 ± 2.1% vs. 21.2 ± 2.1%, p = 0.02) were lower than TomTec measurements while SC4_LS was similar (21.3 ± 2.7% vs. 21.3 ± 2.5%, p = 0.94). Mean differences between EchoPAC and TomTec were ≤ 0.6% strain units for all subcostal LVLS measurements; SCav_LS showed the narrowest limits of agreement (LOA) (mean difference - 0.3%, LOA - 3.2 to 2.6%). EchoPAC and TomTec measurements of SCav_LS showed good correlation (r = 0.76, p < 0.001). Intra-observer and inter-observer analysis showed good reproducibility. Inter-observer variability was lower than inter-vendor variability; SCav_LS was most reproducible: inter-observer relative mean error was 3.6% for EchoPAC and 4.3% for TomTec and inter-observer LOA were ± 2.1% for EchoPAC and ± 2.6% for TomTec. Averaged subcostal LVLS was highly reproducible with inter-observer variability comparable to GLS. Inter-vendor differences in averaged subcostal LVLS were small but statistically significant.

Prospective multicentre cohort study of transthoracic echocardiography provision in the South West of the UK during the first wave of SARS-CoV-2 pandemic

AIMS

It was predicted internationally that transthoracic echocardiography (TTE) would be vital during the SARS-CoV-2 outbreak. We therefore, designed a study to report the demand for TTE in two large District General Hospitals during the rise in the first wave of the SARS-CoV-2 pandemic in the UK. A primary clinical outcome of 30-day mortality was also assessed.

METHODS

The TTE service across two hospitals was reconfigured to maximise access to inpatient scanning. All TTEs of suspected or confirmed SARS-CoV-2 patients over a 3-week period were included in the study. All patients were followed up until at least day 30 after their scan at which point the primary clinical outcome of mortality was recorded. Comparative analysis based on mortality was conducted for all TTE results, biochemical markers and demographics.

RESULTS

27 patients with confirmed SARS-CoV-2 had a TTE within the inclusion window. Mortality comparative analysis showed the deceased group were significantly older (mean 68.4, SD 11.9 vs 60.5, SD 13.0, p=0.03) and more commonly reported fatigue in their presenting symptoms (29.6% vs 71.4%, p=0.01). No other differences were identified in the demographic or biochemical data. Left ventricular systolic dysfunction was noted in 7.4% of patients and right ventricular impairment or dilation was seen in 18.5% patients. TTE results were not significantly different in mortality comparative analysis.

CONCLUSION

This study demonstrates an achievable approach to TTE services when under increased pressure. Data analysis supports the limited available data suggesting right ventricular abnormalities are the most commonly identified echocardiographic change in SARS-CoV-2 patients. No association can be demonstrated between mortality and TTE results.

Feasibility and Reproducibility of Left Atrium Measurements Using Different Three-Dimensional Echocardiographic Modalities

Left atrium (LA) volume is a biomarker of cardiovascular outcomes. Three-dimensional echocardiography (3DE) provides an accurate LA evaluation, but data regarding the optimal 3DE method is scarce. We assessed the feasibility and reproducibility of LA measurements using different 3DE methods. One hundred and ninety-four patients were prospectively analyzed. Conventional 3DE and two semi-automatic 3DE algorithms (Tomtec™ and Dynamic Heart Model (DHM)) were used in 110 patients. Intra- and interobserver reproducibility and intervendor comparison were performed in additional patients' subsets. Forty patients underwent cardiac magnetic resonance (CMR). Feasibility was 100% for Tomtec, 98.2% for DHM, and 72.8% for conventional 3DE. Tomtec volumes were higher than 3DE and DHM (p < 0.001). Reproducibility was better for DHM (intraobserver LA maximum volume (LAmax) ICC 0.99 (95% CI 1.0-0.99), LA minimum volume (LAmin) 0.98 (95% CI 0.95-0.99), LApreA 0.96 (95% CI 0.91-0.98); interobserver LAmax ICC 0.98 (95% CI 0.96-0.99), LAmin 0.99 (95% CI 0.99-1.00), and LApreA 0.97 (95% CI 0.94-0.99)). Intervendor comparison showed differences between left ventricle (LV) software adapted for LA (p < 0.001). Tomtec underestimated the least LA volumes compared to CMR. These findings emphasize that dedicated software should be used for LA assessment, for consistent clinical longitudinal follow-up and research.

Artificial Intelligence Aids Cardiac Image Quality Assessment for Improving Precision in Strain Measurements

OBJECTIVES

The aim of this study was to develop an artificial intelligence tool to assess echocardiographic image quality objectively.

BACKGROUND

Left ventricular global longitudinal strain (LVGLS) has recently been used to monitor cancer therapeutics-related cardiac dysfunction (CTRCD) but image quality limits its reliability.

METHODS

A DenseNet-121 convolutional neural network was developed for view identification from an athlete's echocardiographic dataset. To prove the concept that classification confidence (CC) can serve as a quality marker, values of longitudinal strain derived from feature tracking of cardiac magnetic resonance (CMR) imaging and strain analysis of echocardiography were compared. The CC was then applied to patients with breast cancer free from CTRCD to investigate the effects of image quality on the reliability of strain analysis.

RESULTS

CC of the apical 4-chamber view (A4C) was significantly correlated with the endocardial border delineation index. CC of A4C >900 significantly predicted a <15% relative difference in longitudinal strain between CMR feature tracking and automated echocardiographic analysis. Echocardiographic studies (n =752) of 102 patients with breast cancer without CTRCD were investigated. The strain analysis showed higher parallel forms, inter-rater, and test-retest reliabilities in patients with CC of A4C >900. During sequential comparisons of automated LVGLS in individual patients, those with CC of A4C >900 had a lower false positive detection rate of CTRCD.

CONCLUSIONS

CC of A4C was associated with the reliability of automated LVGLS and could also potentially be used as a filter to select comparable images from sequential echocardiographic studies in individual patients and reduce the false positive detection rate of CTRCD.

Prognostic Value of Early Systolic Lengthening by Strain Imaging in Type 2 Diabetes

BACKGROUND

Patients with type 2 diabetes (T2D) have increased risk for subclinical myocardial disease. Early systolic lengthening (ESL), a paradoxical stretch of myocardial fibers, is a sensitive marker of myocardial dysfunction. The aims of this study were to investigate the prognostic value of ESL in patients with T2D and to determine if global longitudinal strain (GLS) modifies this relationship.

METHODS

In this prospective study, speckle-tracking echocardiography was conducted in 703 patients with T2D (62% men; mean age, 63 ± 10 years; median diabetes duration, 11 years; interquartile range, 6-17 years). Patients had no histories of significant heart disease. ESL index was assessed as [-100 × (peak positive systolic strain/maximal strain)] and ESL duration as time from QRS complex on the electrocardiogram to time of peak positive systolic strain. P values ≤ .004 were considered to indicate statistical significance.

RESULTS

During a median follow-up time of 4.8 years (interquartile range, 4.1-5.3 years), 86 patients (12%) experienced major adverse cardiovascular events (MACE), a composite of incident heart failure, myocardial infarction, and cardiovascular death. In multivariate models, only the ESL index (hazard ratio [HR], 1.06 per 1% increase; 95% CI, 1.01-1.010; P = .004) but not ESL duration (HR, 1.02 per 1-ms increase; 95% CI, 1.00-1.03; P = .036) were associated with MACE. GLS modified this relationship (P for interaction < .05) such that in patients with low GLS (>-18%), ESL index (HR, 1.06 per 1% increase; 95% CI, 1.02-1.10; P = .003) was associated with MACE, but ESL duration was not (HR, 1.02 per 1-ms increase; 95% CI, 1.00-1.04; P = .005). No associations were found for high GLS (<-18%).

CONCLUSIONS

In patients with T2D and no histories of heart disease, ESL provides prognostic information on MACE and may potentially aid in cardiovascular risk stratification.

Reproducibility and Intervendor Agreement of Left Ventricular Global Systolic Strain in Children Using a Layer-Specific Analysis

BACKGROUND

Speckle-tracking strain analysis provides additive data to the assessment of pediatric and congenital heart disease; however, the variety of strain analysis software platforms by different vendors and the lack of data on intervendor strain agreement in children have limited its utility. The purpose of this study is to evaluate the intervendor agreement of strain on two commonly used analysis platforms in pediatrics by layer of myocardium and data compression.

METHODS

This prospective study analyzed two-dimensional speckle-tracking strain on two software platforms in 53 children with normal cardiac segmental anatomy and varying function. Three standard apical views and one parasternal short-axis view were exported at their acquired frame rates to workstations with GE EchoPAC and TomTec software and then also to TomTec at compressed frame rates. Both software platforms had been updated with European Association of Cardiovascular Imaging/American Society of Echocardiography Task Force recommendations for left ventricular (LV) global strain. Intravendor and intervendor agreement between layer-specific comparisons were assessed using Bland-Altman analysis (limits of agreement and bias) and intraclass correlation coefficients.

RESULTS

This study included subjects with normal LV function (n = 38) and cardiomyopathy (n = 15) with an age range of 1 month to 18 years. Intertechnique agreement by default vendor myocardial layer (GE mid-TomTec endocardial layer) was robust for both global longitudinal (GLS) and circumferential strain (GCS; higher for GLS than GCS). Intravendor (inter- and interreader) agreement was slightly higher than intervendor. Only small differences in intraclass correlation coefficients were present between various myocardial layers and acquired versus compressed TomTec data with narrow limits of agreement and small bias except in certain subgroup comparisons.

CONCLUSIONS

Comparison of LV GLS and GCS between two commonly used software platforms after European Association of Cardiovascular Imaging/American Society of Echocardiography Industry Task Force recommendations demonstrated good to excellent agreement in pediatrics, regardless of the layer of analysis or the image format, although some degree of variability remains between vendor platforms. Overall, GLS agreement was more robust than GCS, and this difference is exaggerated in specific subanalyses. These data suggest that comparisons of strain values obtained on these two vendors will be reasonable, but caution should be used when the indication is the detection of small differences between serial echocardiograms.

Accuracy and reliability of novel semi-automated two-dimensional layer specific speckle tracking software for quantifying left ventricular volumes and function

Purpose

To determine whether the semi-automated two-dimensional echocardiography (2DE) layer strain software, compared to cardiac magnetic resonance (CMR), is reliable for left ventricular (LV) volume quantification.

Methods and results

We retrospectively selected 84 patients who underwent CMR and 2DE on the same day. Novel 2DE layer strain software automatically provides LV contour in 3 myocardial layers and performs layer specific speckle tracking analysis, which calculates LV volumes, ejection fraction (LVEF), and global longitudinal strain (GLS) in each layer. These values were compared with reference values from CMR disk-area summation and feature tracking methods. Coverage probability (CP) was determined using predefined cut-off values and absolute differences in LV volumes of 30 mL, those in LVEF of 10%, and those in GLS of 4%. The software did not work in 3 patients (feasibility: 96%). Different layers resulted in different degrees of under- or over-estimation of LV volumes. Epicardial tracking significantly overestimated the LV volumes and significantly underestimated LVEF and GLS. Mid-myocardial tracking had less underestimation of LV volumes and equivalent CP values of LVEF (0.77 vs. 0.75 using the disk-area summation method and 0.90 vs. 0.94 using the feature tracking method) and GLS (0.95 vs. 0.92) compared with endocardial tracking. The new software showed excellent reproducibility, especially endocardial and mid-myocardial tracking.

Conclusions

Mid-myocardial tracking with the novel 2DE strain software provided less bias of LV volumes with high CP values of LVEF and GLS, which suggests that mid-myocardial layer speckle tracking analysis approximates CMR derived LV functional parameters.

Accurate assessment of LV function using the first automated 2D-border detection algorithm for small animals - evaluation and application to models of LV dysfunction

Echocardiography is the most commonly applied technique for non-invasive assessment of cardiac function in small animals. Manual tracing of endocardial borders is time consuming and varies with operator experience. Therefore, we aimed to evaluate a novel automated two-dimensional software algorithm (Auto2DE) for small animals and compare it to the standard use of manual 2D-echocardiographic assessment (2DE). We hypothesized that novel Auto2DE will provide rapid and robust data sets, which are in agreement with manually assessed data of animals.2DE and Auto2DE were carried out using a high-resolution imaging-system for small animals. First, validation cohorts of mouse and rat cine loops were used to compare Auto2DE against 2DE. These data were stratified for image quality by a blinded expert in small animal imaging. Second, we evaluated 2DE and Auto2DE in four mouse models and four rat models with different cardiac pathologies.Automated assessment of LV function by 2DE was faster than conventional 2DE analysis and independent of operator experience levels. The accuracy of Auto2DE-assessed data in healthy mice was dependent on cine loop quality, with excellent agreement between Auto2DE and 2DE in cine loops with adequate quality. Auto2DE allowed for valid detection of impaired cardiac function in animal models with pronounced cardiac phenotypes, but yielded poor performance in diabetic animal models independent of image quality.Auto2DE represents a novel automated analysis tool for rapid assessment of LV function, which is suitable for data acquisition in studies with good and very good echocardiographic image quality, but presents systematic problems in specific pathologies.