Automated Quantification in Echocardiography

Artificial intelligence-enhanced automation for M-mode echocardiographic analysis: ensuring fully automated, reliable, and reproducible measurements

To enhance M-mode echocardiography's utility for measuring cardiac structures, we developed and evaluated an artificial intelligence (AI)-based automated analysis system for M-mode images through the aorta and left atrium [M-mode (Ao-LA)], and through the left ventricle [M-mode (LV)]. Our system, integrating two deep neural networks (DNN) for view classification and image segmentation, alongside an auto-measurement algorithm, was developed using 5,958 M-mode images [3,258 M-mode (LA-Ao), and 2,700 M-mode (LV)] drawn from a nationwide echocardiographic dataset collated from five tertiary hospitals. The performance of view classification and segmentation DNNs were evaluated on 594 M-mode images, while automatic measurement accuracy was tested on separate internal test set with 100 M-mode images as well as external test set with 280 images (140 sinus rhythm and 140 atrial fibrillation). Performance evaluation showed the view classification DNN's overall accuracy of 99.8% and segmentation DNN's Dice similarity coefficient of 94.3%. Within the internal test set, all automated measurements, including LA, Ao, and LV wall and cavity, resonated strongly with expert evaluations, exhibiting Pearson's correlation coefficients (PCCs) of 0.81-0.99. This performance persisted in the external test set for both sinus rhythm (PCC, 0.84-0.98) and atrial fibrillation (PCC, 0.70-0.97). Notably, automatic measurements, consistently offering multi-cardiac cycle readings, showcased a stronger correlation with the averaged multi-cycle manual measurements than with those of a single representative cycle. Our AI-based system for automatic M-mode echocardiographic analysis demonstrated excellent accuracy, reproducibility, and speed. This automated approach has the potential to improve efficiency and reduce variability in clinical practice.

Innovations in Diabetic Macular Edema Management: A Comprehensive Review of Automated Quantification and Anti-vascular Endothelial Growth Factor Intervention

Diabetic macular edema (DME) poses a significant threat to the vision and quality of life of individuals with diabetes. This comprehensive review explores recent advancements in DME management, focusing on integrating automated quantification techniques and anti-vascular endothelial growth factor (anti-VEGF) interventions. The review begins with an overview of DME, emphasizing its prevalence, impact on diabetic patients, and current challenges in management. It then delves into the potential of automated quantification, leveraging machine learning and artificial intelligence to improve early detection and monitoring. Concurrently, the role of anti-VEGF therapies in addressing the underlying vascular abnormalities in DME is scrutinized. The review synthesizes vital findings, highlighting the implications for the future of DME management. Promising outcomes from recent clinical trials and case studies are discussed, providing insights into the evolving landscape of personalized medicine approaches. The conclusion underscores the transformative potential of these innovations, calling for continued research, collaboration, and integration of these advancements into clinical practice. This review aims to serve as a roadmap for researchers, clinicians, and industry stakeholders, fostering a collective effort to enhance the precision and efficacy of DME management.

Factors associated with reporting left ventricular ejection fraction with 3D echocardiography in real-world practice

BACKGROUND

Guidelines recommend 3D echocardiography (3DE) to assess left ventricular ejection fraction (LVEF) on transthoracic echocardiogram (TTE) when possible, but it is unclear which factors are most strongly associated with reporting 3DE LVEF in real-world practice.

METHODS

We evaluated 3DE LVEF reporting by age, sex, BMI, TTE location and variation in reporting by sonographer and reader. All TTEs were performed without contrast enhancement agent at a large medical center from 9/2015 to 12/2020 using ultrasound machines capable of 3DE. We used multivariable logistic regression to assess which factors were most associated with reporting 3DE LVEF.

RESULTS

Among 35 641 TTEs included in this study, 57.4% were performed on women. 3DE LVEF was reported on 18 391 TTEs (51.6% of cohort; 50.5% for women and 52.4% for men). Portable inpatient TTEs (n = 5569) had the lowest rates of 3DE LVEF reporting (30.9%), while general outpatient TTEs (n = 15 933) had greater reporting (56.9%). Outpatient TTEs with an indication for chemotherapy (n = 3244) had the highest rates of 3DE LVEF (87.2%). The median (IQR) percentage of TTEs reporting 3D LVEF was 52.7% (43.1%-68.1%) among sonographers and 51.6% (46.5%-59.6%) among readers. Among 20082 (56.3%) TTEs with 3DE LVEF measured by sonographers, 91.6% were included by readers in the final report. After adjustment, performing sonographer in the highest reporting quartile was most strongly associated with reporting 3DE LVEF (OR 7.04, 95% CI 6.55-7.56), while an inpatient portable study had the strongest negative association for reporting (OR .38, 95% CI .35-.40).

CONCLUSIONS

Use of 3DE LVEF in real-world practice varies substantially based on performing sonographer and is low for hospitalized patients, but can be frequently used for chemotherapy. Initiatives are needed to increase sonographer 3DE acquisition in most clinical settings.

Quantification of Myocardial Contraction Fraction with Three-Dimensional Automated, Machine-Learning-Based Left-Heart-Chamber Metrics: Diagnostic Utility in Hypertrophic Phenotypes and Normal Ejection Fraction

Aims: The differentiation of left ventricular (LV) hypertrophic phenotypes is challenging in patients with normal ejection fraction (EF). The myocardial contraction fraction (MCF) is a simple dimensionless index useful for specifically identifying cardiac amyloidosis (CA) and hypertrophic cardiomyopathy (HCM) when calculated by cardiac magnetic resonance. The purpose of this study was to evaluate the value of MCF measured by three-dimensional automated, machine-learning-based LV chamber metrics (dynamic heart model [DHM]) for the discrimination of different forms of hypertrophic phenotypes. Methods and Results: We analyzed the DHM LV metrics of patients with CA (n = 10), hypertrophic cardiomyopathy (HCM, n = 36), isolated hypertension (IH, n = 87), and 54 healthy controls. MCF was calculated by dividing LV stroke volume by LV myocardial volume. Compared with controls (median 61.95%, interquartile range 55.43-67.79%), mean values for MCF were significantly reduced in HCM-48.55% (43.46-54.86% p < 0.001)-and CA-40.92% (36.68-46.84% p < 0.002)-but not in IH-59.35% (53.22-64.93% p < 0.7). MCF showed a weak correlation with EF in the overall cohort (R2 = 0.136) and the four study subgroups (healthy adults, R2 = 0.039 IH, R2 = 0.089; HCM, R2 = 0.225; CA, R2 = 0.102). ROC analyses showed that MCF could differentiate between healthy adults and HCM (sensitivity 75.9%, specificity 77.8%, AUC 0.814) and between healthy adults and CA (sensitivity 87.0%, specificity 100%, AUC 0.959). The best cut-off values were 55.3% and 52.8%. Conclusions: The easily derived quantification of MCF by DHM can refine our echocardiographic discrimination capacity in patients with hypertrophic phenotype and normal EF. It should be added to the diagnostic workup of these patients.

Quantification of left ventricular ejection fraction and cardiac output using a novel semi-automated echocardiographic method: a prospective observational study in coronary artery bypass patients

BACKGROUND

Echocardiographic quantification of ejection fraction (EF) by manual endocardial tracing requires training, is time-consuming and potentially user-dependent, whereas determination of cardiac output by pulmonary artery catheterization (PAC) is invasive and carries a risk of complications. Recently, a novel software for semi-automated EF and CO assessment (AutoEF) using transthoracic echocardiography (TTE) has been introduced. We hypothesized that AutoEF would provide EF values different from those obtained by the modified Simpson's method in transoesophageal echocardiography (TOE) and that AutoEF CO measurements would not agree with those obtained via VTI_LVOT in TOE and by thermodilution using PAC.

METHODS

In 167 patients undergoing coronary artery bypass graft surgery (CABG), TTE cine loops of apical 4- and 2-chamber views were recorded after anaesthesia induction under steady-state conditions. Subsequently, TOE was performed following a standardized protocol, and CO was determined by thermodilution. EF and CO were assessed by TTE AutoEF as well as TOE, using the modified Simpson's method, and Doppler measurements via velocity time integral in the LV outflow tract (VTI_LVOT). We determined Pearson's correlation coefficients r and carried out Bland-Altman analyses. The primary endpoints were differences in EF and CO. The secondary endpoints were differences in left ventricular volumes at end diastole (LVEDV) and end systole (LVESV).

RESULTS

AutoEF and the modified Simpson's method in TOE showed moderate EF correlation (r = 0.38, p < 0.01) with a bias of -12.6% (95% limits of agreement (95%LOA): -36.6 - 11.3%). AutoEF CO correlated poorly both with VTI_LVOT in TOE (r = 0.19, p < 0.01) and thermodilution (r = 0.28, p < 0.01). The CO bias between AutoEF and VTI_LVOT was 1.33 l min^-1 (95%LOA: -1.72 - 4.38 l min^-1) and 1.39 l min^-1 (95%LOA -1.34 - 4.12 l min^-1) between AutoEF and thermodilution, respectively. AutoEF yielded both significantly lower EF (EF_AutoEF: 42.0% (IQR 29.0 - 55.0%) vs. EF_{TOE Simpson}: 55.2% (IQR 40.1 - 70.3%), p < 0.01) and CO values than the reference methods (CO_{AutoEF biplane}: 2.30 l min^-1 (IQR 1.30 - 3.30 l min^-1) vs. CO_{VTI LVOT}: 3.64 l min^-1 (IQR 2.05 - 5.23 l min^-1) and CO_PAC: 3.90 l min^-1 (IQR 2.30 - 5.50 l min^-1), p < 0.01)).

CONCLUSIONS

AutoEF correlated moderately with TOE EF determined by the modified Simpson's method but poorly both with VTI_LVOT and thermodilution CO. A systematic bias was detected overestimating LV volumes and underestimating both EF and CO compared to the reference methods.

TRIAL REGISTRATION

German Register for Clinical Trials (DRKS-ID DRKS00010666, date of registration: 08/07/2016).

Assisted probe guidance in cardiac ultrasound: A review

Echocardiography is the most frequently used imaging modality in cardiology. However, its acquisition is affected by inter-observer variability and largely dependent on the operator's experience. In this context, artificial intelligence techniques could reduce these variabilities and provide a user independent system. In recent years, machine learning (ML) algorithms have been used in echocardiography to automate echocardiographic acquisition. This review focuses on the state-of-the-art studies that use ML to automate tasks regarding the acquisition of echocardiograms, including quality assessment (QA), recognition of cardiac views and assisted probe guidance during the scanning process. The results indicate that performance of automated acquisition was overall good, but most studies lack variability in their datasets. From our comprehensive review, we believe automated acquisition has the potential not only to improve accuracy of diagnosis, but also help novice operators build expertise and facilitate point of care healthcare in medically underserved areas.

Three-Dimensional Automated, Machine-Learning-Based Left Heart Chamber Metrics: Associations with Prevalent Vascular Risk Factors and Cardiovascular Diseases

Background. Three-dimensional transthoracic echocardiography (3DE) powered by artificial intelligence provides accurate left chamber quantification in good accordance with cardiac magnetic resonance and has the potential to revolutionize our clinical practice. Aims. To evaluate the association and the independent value of dynamic heart model (DHM)-derived left atrial (LA) and left ventricular (LV) metrics with prevalent vascular risk factors (VRFs) and cardiovascular diseases (CVDs) in a large, unselected population. Materials and Methods. We estimated the association of DHM metrics with VRFs (hypertension, diabetes) and CVDs (atrial fibrillation, stroke, ischemic heart disease, cardiomyopathies, >moderate valvular heart disease/prosthesis), stratified by prevalent disease status: participants without VRFs or CVDs (healthy), with at least one VRFs but without CVDs, and with at least one CVDs. Results. We retrospectively included 1069 subjects (median age 62 [IQR 49−74]; 50.6% women). When comparing VRFs with the healthy, significant difference in maximum and minimum indexed atrial volume (LAVi max and LAVi min), left atrial ejection fraction (LAEF), left ventricular mass/left ventricular end-diastolic volume ratio, and left ventricular global function index (LVGFI) were recorded (p < 0.05). In the adjusted logistic regression, LAVi min, LAEF, LV ejection fraction, and LVGFI showed the most robust association (OR 3.03 [95% CI 2.48−3.70], 0.45 [95% CI 0.39−0.51], 0.28 [95% CI 0.22−0.35], and 0.22 [95% CI 0.16−0.28], respectively, with CVDs. Conclusions. The present data suggested that novel 3DE left heart chamber metrics by DHM such as LAEF, LAVi min, and LVGFI can refine our echocardiographic disease discrimination capacity.

A multiparameter algorithm to guide repair of degenerative mitral regurgitation

PURPOSE

Degenerative mitral regurgitation repair using a measured algorithm could increase the precision and reproducibility of repair outcomes.

METHODS

Direct and echocardiographic measurements guide the repair to achieve a coaptation length of 5 to 10 mm and minimize the risk of systolic anterior motion. Leaflet reconstruction restored the normal 2 to 1 ratio of anterior to posterior leaflet length without residual prolapse or restriction. The choice of ring size was based on anterior leaflet length, the distance from the leaflet coaptation point to the septum, and the anterior-posterior ring dimension. Freedom from reoperation and mitral regurgitation recurrence were based on multistate models.

RESULTS

One thousand fifty-one patients had mitral surgery and 1026 (97.6%) were repaired. A2 length was 27.2 ± 4.5 mm; and the reconstructed posterior leaflet was 13.9 ± 2.3 mm. Median ring size was 34 mm and strongly correlated to A2 length (R = 0.76; P < .001). The coaptation length at P2 after repair was 6.4 ± 1.7 mm and 87% of measurements were between 5 and 10 mm. Results at predischarge and 10 years, respectively, included mild regurgitation (7.5% and 26.1%), moderate (0.7% and 15.6%), moderate to severe (0% and 1.4%), and severe (0% and 0%), with mean mitral gradient values 3.5 ± 1.5 and 2.9 ± 1.2 mm Hg, respectively. Systolic anterior motion at discharge and last follow-up were 0.2% and 1.1%, respectively. Ten-year freedom from mitral valve reoperation was 99.7%.

CONCLUSIONS

A simple, reproducible, measured algorithm for degenerative mitral valve repair provides excellent early and late results and is a useful adjunct to established surgical techniques.

Real-time echocardiography image analysis and quantification of cardiac indices

Deep learning has a huge potential to transform echocardiography in clinical practice and point of care ultrasound testing by providing real-time analysis of cardiac structure and function. Automated echocardiography analysis is benefited through use of machine learning for tasks such as image quality assessment, view classification, cardiac region segmentation, and quantification of diagnostic indices. By taking advantage of high-performing deep neural networks, we propose a novel and eicient real-time system for echocardiography analysis and quantification. Our system uses a self-supervised modality-specific representation trained using a publicly available large-scale dataset. The trained representation is used to enhance the learning of target echo tasks with relatively small datasets. We also present a novel Trilateral Attention Network (TaNet) for real-time cardiac region segmentation. The proposed network uses a module for region localization and three lightweight pathways for encoding rich low-level, textural, and high-level features. Feature embeddings from these individual pathways are then aggregated for cardiac region segmentation. This network is fine-tuned using a joint loss function and training strategy. We extensively evaluate the proposed system and its components, which are echo view retrieval, cardiac segmentation, and quantification, using four echocardiography datasets. Our experimental results show a consistent improvement in the performance of echocardiography analysis tasks with enhanced computational eiciency that charts a path toward its adoption in clinical practice. Specifically, our results show superior real-time performance in retrieving good quality echo from individual cardiac view, segmenting cardiac chambers with complex overlaps, and extracting cardiac indices that highly agree with the experts’ values. The source code of our implementation can be found in the project ‘ s GitHub page.

Advanced Echocardiographic Techniques in Cardio-Oncology: the Role for Early Detection of Cardiotoxicity

PURPOSE OF REVIEW

Implementation of advanced echocardiographic techniques in cardio-oncology is a growing need as they are the cornerstone of early detection of cancer therapy-related cardiovascular toxicity (CTR-CVT).

RECENT FINDINGS

Three-dimensional echocardiography and myocardial deformation techniques have shown more accuracy and reproducibility than classic 2D measurements in detecting cardiovascular adverse effects in patients undergoing anticancer therapies. Application of advanced echo techniques to daily monitoring of patients with cancer helps to identify those at risk of developing CTR-CVT during and after cancer treatment. Furthermore, advanced echo parameters improve early initiation of cardioprotective treatments in order to minimize cardiovascular events and cancer treatment interruption.

Elucidating tricuspid Doppler signal interpolation and its implication for assessing pulmonary hypertension

Doppler echocardiography plays a central role in the assessment of pulmonary hypertension (PAH). We aim to improve quality assessment of systolic pulmonary arterial pressure (SPAP) by applying a cubic polynomial interpolation to digitized tricuspid regurgitation (TR) waveforms. Patients with PAH and advanced lung disease were divided into three cohorts: a derivation cohort (n = 44), a validation cohort (n = 71), an outlier cohort (n = 26), and a non-PAH cohort (n = 44). We digitized TR waveforms and analyzed normalized duration, skewness, kurtosis, and first and second derivatives of pressure. Cubic polynomial interpolation was applied to three physiology-driven phases: the isovolumic phase, ejection phase, and "shoulder" point phase. Coefficients of determination and a Bland-Altman analysis was used to assess bias between methods. The cubic polynomial interpolation of the TR waveform correlated strongly with expert read right ventricular systolic pressure (RVSP) with R 2 > 0.910 in the validation cohort. The biases when compared to invasive SPAP measured within 24 h were 6.03 [4.33; 7.73], -2.94 [1.47; 4.41], and -3.11 [-4.52; -1.71] mmHg, for isovolumic, ejection, and shoulder point interpolations, respectively. In the outlier cohort with more than 30% difference between echocardiographic estimates and invasive SPAP, cubic polynomial interpolation significantly reduced underestimation of RVSP. Cubic polynomial interpolation of the TR waveform based on isovolumic or early ejection phase may improve RVSP estimates.

Reliability and feasibility of automated function imaging for quantification in patients with left ventricular dilation: comparison with cardiac magnetic resonance

Automated function imaging (AFI, GE Healthcare) is a novel promising algorithm of speckle-tracking echocardiography that combines two-dimensional strain and AI technology. It shortens the analysis time, saves the cost associated with streamlining of image acquisition, rapid analysis, and reporting, and has greater accuracy and reproducibility of measurements. This study aimed to evaluate the reliability and feasibility of AFI for the quantification of left ventricular (LV) volumes and function in patients with LV dilation by comparison with CMR. We retrospectively studied 50 patients with LV dilation on echocardiography whom both underwent CMR and coronary angiography within three days. LV volumes, ejection fraction (EF), and global longitudinal strain (GLS) were measured from 3 long-axis cine-views via the AFI technique in two modes: without editing (auto-AFI) and with partial border editing (semi-auto-AFI). The LV volumes and EF were also measured with 2D Simpson's biplane method, and CMR, as the standard method, was used for comparison. The AFI method still had significantly underestimated the LV volumes compared with CMR (P<0.01), but there were no significant differences between the AFI method and the conventional Simpson's biplane method. There were no significant differences in EF between CMR and the AFI method with good correlations (auto-AFI: r = 0.81, semi-auto-AFI: r = 0.86). The auto-AFI method provided the most rapid analysis and excellent reproducibility, while the semi-auto-AFI method further improved measurement accuracy. However, there were no significant differences in LV volumes and EF between these two AFI methods. The accuracy of AFI seems to be more affected by the image quality than the left ventricular morphology.  AFI enables accurate, efficient, and rapid evaluation of LV volumes and EF in patients with dilated LV, with good reproducibility and correlations with CMR.

An integrated approach to cardioprotection in lymphomas

In potentially curable cancers, long-term survival depends not only on the successful treatment of the malignancy but also on the risks associated with treatment-related toxicity, especially cardiotoxicity. Malignant lymphomas affect patients at any age, with acute and late toxicity risks that could have a severe effect on morbidity, mortality, and quality of life. Although our understanding of chemotherapy-associated and radiotherapy-associated cardiovascular disease has advanced considerably, new drugs with potential cardiotoxicity have been introduced for the treatment of lymphomas. In this Review, we summarise the mechanisms of treatment-related cardiac injury, available clinical data, and protocols for optimising cardioprotection in lymphomas. We discuss ongoing research strategies to advance our knowledge of the molecular basis of drug-induced and radiation-induced toxicity. Additionally, we emphasise the potential for personalised follow-up and early detection, including the role of biomarkers and novel diagnostic tests, highlighting the role of the cardio-oncology team.

Artificial Intelligence-Enhanced Echocardiography for Systolic Function Assessment

The accurate assessment of left ventricular systolic function is crucial in the diagnosis and treatment of cardiovascular diseases. Left ventricular ejection fraction (LVEF) and global longitudinal strain (GLS) are the most critical indexes of cardiac systolic function. Echocardiography has become the mainstay of cardiac imaging for measuring LVEF and GLS because it is non-invasive, radiation-free, and allows for bedside operation and real-time processing. However, the human assessment of cardiac function depends on the sonographer’s experience, and despite their years of training, inter-observer variability exists. In addition, GLS requires post-processing, which is time consuming and shows variability across different devices. Researchers have turned to artificial intelligence (AI) to address these challenges. The powerful learning capabilities of AI enable feature extraction, which helps to achieve accurate identification of cardiac structures and reliable estimation of the ventricular volume and myocardial motion. Hence, the automatic output of systolic function indexes can be achieved based on echocardiographic images. This review attempts to thoroughly explain the latest progress of AI in assessing left ventricular systolic function and differential diagnosis of heart diseases by echocardiography and discusses the challenges and promises of this new field.

Prognostic value of diastolic function parameters in significant aortic regurgitation: the role of the left atrial strain

BACKGROUND

The management of patients with asymptomatic significant aortic regurgitation (sAR) is often challenging and appropriate timing of aortic valve surgery remains controversial. Prognostic value of diastolic parameters has been demonstrated in several cardiac diseases. The aim of this study was to analyze the prognostic significance of the diastolic function evaluated by echocardiography, in asymptomatic patients with sAR.

METHODS

A total of 126 patients with asymptomatic sAR evaluated in the Heart Valve Clinic were retrospective included. Conventional echocardiographic systolic and diastolic function parameters were assessed. Left atrial (LA) auto-strain analysis was performed in a sub-group of 57 patients. A combined end-point of hospital admission due to heart failure, cardiovascular mortality, or aortic valve surgery was defined.

RESULTS

During a median follow-up of 34.1 (interquartile range 16.5-48.1) months, 25 (19.8%) patients reached the combined end-point. Univariate analysis showed that LV volumes, LV ejection fraction (LVEF), LV-GLS, E wave, E/e' ratio, LA volume and LA reservoir strain (LASr) were significant predictors of events. Multivariate analysis that tested all classical echocardiographic variables statistically significant in the univariate model showed that LVEDV (HR = 1.02; 95% CI 1.01-1.03; p < 0.001) and E/e' ratio (HR = 1.12; 95% CI 1.03-123; p = 0.01) were significant predictors of events. Kaplan-Meier curve, stratified by median value of LASr, showed that lower LASr values (less than median of 34%) were associated with higher rates of events (p = 0.013).

CONCLUSION

In this population of asymptomatic patients with sAR and normal LV systolic function, baseline diastolic parameters were prognostic markers of cardiovascular events; among them, LASr played a significant predictor role.

Fully automated quantification of cardiac chamber and function assessment in 2-D echocardiography: clinical feasibility of deep learning-based algorithms

We aimed to compare the segmentation performance of the current prominent deep learning (DL) algorithms with ground-truth segmentations and to validate the reproducibility of the manually created 2D echocardiographic four cardiac chamber ground-truth annotation. Recently emerged DL based fully-automated chamber segmentation and function assessment methods have shown great potential for future application in aiding image acquisition, quantification, and suggestion for diagnosis. However, the performance of current DL algorithms have not previously been compared with each other. In addition, the reproducibility of ground-truth annotations which are the basis of these algorithms have not yet been fully validated. We retrospectively enrolled 500 consecutive patients who underwent transthoracic echocardiogram (TTE) from December 2019 to December 2020. Simple U-net, Res-U-net, and Dense-U-net algorithms were compared for the segmentation performances and clinical indices such as left atrial volume (LAV), left ventricular end diastolic volume (LVEDV), left ventricular end systolic volume (LVESV), LV mass, and ejection fraction (EF) were evaluated. The inter- and intra-observer variability analysis was performed by two expert sonographers for a randomly selected echocardiographic view in 100 patients (apical 2-chamber, apical 4-chamber, and parasternal short axis views). The overall performance of all DL methods was excellent [average dice similarity coefficient (DSC) 0.91 to 0.95 and average Intersection over union (IOU) 0.83 to 0.90], with the exception of LV wall area on PSAX view (average DSC of 0.83, IOU 0.72). In addition, there were no significant difference in clinical indices between ground truth and automated DL measurements. For inter- and intra-observer variability analysis, the overall intra observer reproducibility was excellent: LAV (ICC = 0.995), LVEDV (ICC = 0.996), LVESV (ICC = 0.997), LV mass (ICC = 0.991) and EF (ICC = 0.984). The inter-observer reproducibility was slightly lower as compared to intraobserver agreement: LAV (ICC = 0.976), LVEDV (ICC = 0.982), LVESV (ICC = 0.970), LV mass (ICC = 0.971), and EF (ICC = 0.899). The three current prominent DL-based fully automated methods are able to reliably perform four-chamber segmentation and quantification of clinical indices. Furthermore, we were able to validate the four cardiac chamber ground-truth annotation and demonstrate an overall excellent reproducibility, but still with some degree of inter-observer variability.

Artificial intelligence for the echocardiographic assessment of valvular heart disease

Developments in artificial intelligence (AI) have led to an explosion of studies exploring its application to cardiovascular medicine. Due to the need for training and expertise, one area where AI could be impactful would be in the diagnosis and management of valvular heart disease. This is because AI can be applied to the multitude of data generated from clinical assessments, imaging and biochemical testing during the care of the patient. In the area of valvular heart disease, the focus of AI has been on the echocardiographic assessment and phenotyping of patient populations to identify high-risk groups. AI can assist image acquisition, view identification for review, and segmentation of valve and cardiac structures for automated analysis. Using image recognition algorithms, aortic and mitral valve disease states have been directly detected from the images themselves. Measurements obtained during echocardiographic valvular assessment have been integrated with other clinical data to identify novel aortic valve disease subgroups and describe new predictors of aortic valve disease progression. In the future, AI could integrate echocardiographic parameters with other clinical data for precision medical management of patients with valvular heart disease.

Clinical validation of an artificial intelligence-assisted algorithm for automated quantification of left ventricular ejection fraction in real time by a novel handheld ultrasound device

Aims

We sought to evaluate the reliability and diagnostic accuracy of a novel handheld ultrasound device (HUD) with artificial intelligence (AI) assisted algorithm to automatically calculate ejection fraction (autoEF) in a real-world patient population.

Methods and results

We studied 100 consecutive patients (57 ± 15 years old, 61% male), including 38 with abnormal left ventricular (LV) function [LV ejection fraction (LVEF) < 50%]. The autoEF results acquired using the HUD were independently compared with manually traced biplane Simpson's rule measurements on cart-based systems to assess method agreement using intra-class correlation coefficient (ICC), linear regression analysis, and Bland-Altman analysis. The diagnostic accuracy for the detection of LVEF <50% was also calculated. Test-retest reliability of measured EF by the HUD was assessed by calculating the ICC and the minimal detectable change (MDC). The ICC, linear regression analysis, and Bland-Altman analysis revealed good agreement between autoEF and reference manual EF (ICC = 0.85; r = 0.87, P < 0.001; mean bias -1.42% with limits of agreement 14.5%, respectively). Detection of abnormal LV function (EF < 50%) by autoEF algorithm was feasible with sensitivity 90% (95% CI 75-97%), specificity 87% (95% CI 76-94%), PPV 81% (95% CI 66-91%), NPV 93% (95% CI 83-98%), and a total diagnostic accuracy of 88%. Test-retest reliability was excellent (ICC = 0.91, P < 0.001; r = 0.91, P < 0.001; mean difference ± SD: 0.54% ± 5.27%, P = 0.308) and MDC for LVEF measurement by autoEF was calculated at 4.38%.

Conclusion

Use of a novel HUD with AI-enabled capabilities provided similar LVEF results with those derived by manual biplane Simpson's method on cart-based systems and shows clinical potential.

From secondary to tertiary mitral regurgitation: the paradigm shifts, but uncertainties remain

Secondary mitral regurgitation (MR) is the most common and undertreated form of MR, whose contribution to poor prognosis and indications to correction remains under discussion. MR has been characterized into 'proportionate' or 'disproportionate', based on left ventricle (LV) and regurgitant volumes, whereas 'tertiary' MR identifies conditions, in which regurgitation is pathologic per se and actively contributes to LV dysfunction. Echocardiographic and anatomo-pathological studies revealed that secondary MR prompts subtle leaflet maladaptive changes, actively contributing to the dynamic progression of secondary MR. We critically discuss the paradigm shift from secondary to tertiary MR and question the notion that MV leaflets play a passive role in secondary MR. We also review the role of standard transthoracic echocardiography for appraising and quantifying maladaptive MV leaflet changes and LV volumes and call for a more sophisticated and comprehensive imaging framework for classifying MR in future interventional studies.

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.

Aortic Annular Sizing Using Novel Software in Three-Dimensional Transesophageal Echocardiography for Transcatheter Aortic Valve Replacement: A Systematic Review and Meta-Analysis

(1) Background: We performed this study to evaluate the agreement between novel automated software of three-dimensional transesophageal echocardiography (3D-TEE) and multidetector computed tomography (MDCT) for aortic annular measurements of preprocedural transcatheter aortic valve replacement (TAVR); (2) Methods: PubMed, EMBASE, Web of Science, and Cochrane Library (Wiley) databases were systematically searched for studies that compared 3D-TEE and MDCT as the reference standard for aortic annular measurement of the following parameters: annular area, annular perimeter, area derived-diameter, perimeter derived-diameter, maximum and minimum diameter. Meta-analytic methods were utilized to determine the pooled correlations and mean differences between 3D-TEE and MDCT. Heterogeneity and publication bias were also assessed. Meta-regression analyses were performed based on the potential factors affecting the correlation of aortic annular area; (3) Results: A total of 889 patients from 10 studies were included in the meta-analysis. Pooled correlation coefficients between 3D-TEE and MDCT of annulus area, perimeter, area derived-diameter, perimeter derived-diameter, maximum and minimum diameter measurements were strong 0.89 (95% CI: 0.84–0.92), 0.88 (95% CI: 0.83–0.92), 0.87 (95% CI: 0.77–0.93), 0.87 (95% CI: 0.77–0.93), 0.79 (95% CI: 0.64–0.87), and 0.75 (95% CI: 0.61–0.84) (Overall p < 0.0001), respectively. Pooled mean differences between 3D-TEE and MDCT of annulus area, perimeter, area derived-diameter, perimeter derived-diameter, maximum and minimum diameter measurements were −20.01 mm² ((95% CI: −35.37 to −0.64), p = 0.011), −2.31 mm ((95% CI: −3.31 to −1.31), p < 0.0001), −0.22 mm ((95% CI: −0.73 to 0.29), p = 0.40), −0.47 mm ((95% CI: −1.06 to 0.12), p = 0.12), −1.36 mm ((95% CI: −2.43 to −0.30), p = 0.012), and 0.31 mm ((95% CI: −0.15 to 0.77), p = 0.18), respectively. There were no statistically significant associations with the baseline patient characteristics of sex, age, left ventricular ejection fraction, mean transaortic gradient, and aortic valve area to the correlation between 3D-TEE and MDCT for aortic annular area sizing; (4) Conclusions: The present study implies that 3D-TEE using novel software tools, automatically analysis, is feasible to MDCT for annulus sizing in clinical practice.

Echocardiographic Left Ventricular Mass Assessment: Correlation between 2D-Derived Linear Dimensions and 3-Dimensional Automated, Machine Learning-Based Methods in Unselected Patients

A recently developed algorithm for 3D analysis based on machine learning (ML) principles detects left ventricular (LV) mass without any human interaction. We retrospectively studied the correlation between 2D-derived linear dimensions using the ASE/EACVI-recommended formula and 3D automated, ML-based methods (Philips HeartModel) regarding LV mass quantification in unselected patients undergoing echocardiography. We included 130 patients (mean age 60 ± 18 years; 45% women). There was only discrete agreement between 2D and 3D measurements of LV mass (r = 0.662, r² = 0.348, p < 0.001). The automated algorithm yielded an overestimation of LV mass compared to the linear method (Bland–Altman positive bias of 13.1 g with 95% limits of the agreement at 4.5 to 21.6 g, p = 0.003, ICC 0.78 (95%CI 0.68−8.4). There was a significant proportional bias (Beta −0.22, t = −2.9) p = 0.005, the variance of the difference varied across the range of LV mass. When the published cut-offs for LV mass abnormality were used, the observed proportion of overall agreement was 77% (kappa = 0.32, p < 0.001). In consecutive patients undergoing echocardiography for any indications, LV mass assessment by 3D analysis using a novel ML-based algorithm showed systematic differences and wide limits of agreements compared with quantification by ASE/EACVI- recommended formula when the current cut-offs and partition values were applied.

Artificial Intelligence and Machine Learning in Cardiovascular Imaging

Cardiovascular disease is the leading cause of mortality in Western countries and leads to a spectrum of complications that can complicate patient management. The emergence of artificial intelligence (AI) has garnered significant interest in many industries, and the field of cardiovascular imaging is no exception. Machine learning (ML) especially is showing significant promise in various diagnostic imaging modalities. As conventional statistics are reaching their apex in computational capabilities, ML can explore new possibilities and unravel hidden relationships. This will have a positive impact on diagnosis and prognosis for cardiovascular imaging. In this in-depth review, we highlight the role of AI and ML for various cardiovascular imaging modalities.

The Role of Artificial Intelligence in Cardiovascular Imaging: State of the Art Review

In this current digital landscape, artificial intelligence (AI) has established itself as a powerful tool in the commercial industry and is an evolving technology in healthcare. Cutting-edge imaging modalities outputting multi-dimensional data are becoming increasingly complex. In this era of data explosion, the field of cardiovascular imaging is undergoing a paradigm shift toward machine learning (ML) driven platforms. These diverse algorithms can seamlessly analyze information and automate a range of tasks. In this review article, we explore the role of ML in the field of cardiovascular imaging.