The importance of patient characteristics, operators, and image quality for the accuracy of heart failure diagnosis by general practitioners using handheld ultrasound devices.

Aims: To evaluate whether the characteristics of patients, operators, and image quality could explain the accuracy of heart failure (HF) diagnostics by general practitioners (GPs) using handheld ultrasound devices (HUDs) with automatic decision-support software and telemedical support. Methods and results: Patients referred to an outpatient cardiac clinic due to symptoms indicating HF were examined by one of five GPs after dedicated training. In total, 166 patients were included [median (inter-quartile range) age 73 (63-78) years; mean ± standard deviation ejection fraction 53 ± 10%]. The GPs considered whether the patients had HF in four diagnostic steps: (i) clinical examination, (ii) adding focused cardiac HUD examination, (iii) adding automatic decision-support software measuring mitral annular plane systolic excursion (autoMAPSE) and ejection fraction (autoEF), and (iv) adding telemedical support. Overall, the characteristics of patients, operators, and image quality explained little of the diagnostic accuracy. Except for atrial fibrillation [lower accuracy for HUD alone and after adding autoEF (P < 0.05)], no patient characteristics influenced the accuracy. Some differences between operators were found after adding autoMAPSE (P < 0.05). Acquisition errors of the four-chamber view and a poor visualization of the mitral plane were associated with reduced accuracy after telemedical support (P < 0.05). Conclusion: The characteristics of patients, operators, and image quality explained just minor parts of the modest accuracy of GPs' HF diagnostics using HUDs with and without decision-support software. Atrial fibrillation and not well-standardized recordings challenged the diagnostic accuracy. However, the accuracy was only modest in well-recorded images, indicating a need for refinement of the technology.

Cardiac Valve Event Timing in Echocardiography Using Deep Learning and Triplane Recordings.

Cardiac valve event timing plays a crucial role when conducting clinical measurements using echocardiography. However, established automated approaches are limited by the need of external electrocardiogram sensors, and manual measurements often rely on timing from different cardiac cycles. Recent methods have applied deep learning to cardiac timing, but they have mainly been restricted to only detecting two key time points, namely end-diastole (ED) and end-systole (ES). In this work, we propose a deep learning approach that leverages triplane recordings to enhance detection of valve events in echocardiography. Our method demonstrates improved performance detecting six different events, including valve events conventionally associated with ED and ES. Of all events, we achieve an average absolute frame difference (aFD) of maximum 1.4 frames (29 ms) for start of diastasis, down to 0.6 frames (12 ms) for mitral valve opening when performing a ten-fold cross-validation with test splits on triplane data from 240 patients. On an external independent test consisting of apical long-axis data from 180 other patients, the worst performing event detection had an aFD of 1.8 (30 ms). The proposed approach has the potential to significantly impact clinical practice by enabling more accurate, rapid and comprehensive event detection, leading to improved clinical measurements.

Mitral Valve Segmentation and Tracking from Transthoracic Echocardiography Using Deep Learning.

OBJECTIVE: Valvular heart diseases (VHDs) pose a significant public health burden, and deciding the best treatment strategy necessitates accurate assessment of heart valve function. Transthoracic echocardiography (TTE) is the key modality to evaluate VHDs, but the lack of standardized quantitative measurements leads to subjective and time-consuming assessments. We aimed to use deep learning to automate the extraction of mitral valve (MV) leaflets and annular hinge points from echocardiograms of the MV, improving standardization and reducing workload in quantitative assessment of MV disease. METHODS: We annotated the MV leaflets and annulus points in 2931 images from 127 patients. We propose an approach for segmenting the annotated features using Attention UNet with deep supervision and weight scheduling of the attention coefficients to enforce saliency surrounding the MV. The derived segmentation masks were used to extract quantitative biomarkers for specific MV leaflet scallops throughout the heart cycle. RESULTS: Evaluation performance was summarized using a Dice score of 0.63 ± 0.14, annulus error of 3.64 ± 2.53 and leaflet angle error of 8.7 ± 8.3°. Leveraging Attention UNet with deep supervision robustness of clinically relevant metrics was improved compared with UNet, reducing standard deviations by 2.7° (angle error) and 0.73 mm (annulus error). We correctly identified cases of MV prolapse, cases of stenosis and healthy references from a clinical material using the derived biomarkers. CONCLUSION: Robust deep learning segmentation and tracking of MV morphology and motion is possible by leveraging attention gates and deep supervision, and holds promise for enhancing VHD diagnosis and treatment monitoring.

Automated Measurement of Ovary Development in Atlantic Salmon Using Deep Learning.

OBJECTIVE: Salmon breeding companies control the egg stripping period through environmental change, which triggers the need to identify the state of maturation. Ultrasound imaging of the salmon ovary is a proven non-invasive tool for this purpose; however, the process is laborious, and the interpretation of the ultrasound scans is subjective. Real-time ultrasound image segmentation of Atlantic salmon ovary provides an opportunity to overcome these limitations. However, several application challenges need to be addressed to achieve this goal. These challenges include the potential for false-positive and false-negative predictions, accurate prediction of attenuated lower ovary parts and resolution of inconsistencies in predicted ovary shape. METHODS: We describe an approach designed to tackle these obstacles by employing targeted pre-training of a modified U-Net, capable of performing both segmentation and classification. In addition, a variational autoencoder (VAE) and generative adversarial network (GAN) were incorporated to rectify shape inconsistencies in the segmentation output. To train the proposed model, a data set of Atlantic salmon ovaries throughout two maturation periods was recorded. RESULTS: We then tested our model and compared its performance with that of conventional and novel U-Nets. The method was also tested in a salmon on-site ultrasound examination setting. The results of our application indicate that our method is able to efficiently segment salmon ovary with an average Dice score of 0.885 per individual in real-time. CONCLUSION: These results represent a competitive performance for this specific application, which enables us to design an automated system for smart monitoring of maturation state in Atlantic salmon.

Automated Segmentation and Quantification of the Right Ventricle in 2-D Echocardiography.

OBJECTIVE: The right ventricle receives less attention than its left counterpart in echocardiography research, practice and development of automated solutions. In the work described here, we sought to determine that the deep learning methods for automated segmentation of the left ventricle in 2-D echocardiograms are also valid for the right ventricle. Additionally, here we describe and explore a keypoint detection approach to segmentation that guards against erratic behavior often displayed by segmentation models. METHODS: We used a data set of echo images focused on the right ventricle from 250 participants to train and evaluate several deep learning models for segmentation and keypoint detection. We propose a compact architecture (U-Net KP) employing the latter approach. The architecture is designed to balance high speed with accuracy and robustness. RESULTS: All featured models achieved segmentation accuracy close to the inter-observer variability. When computing the metrics of right ventricular systolic function from contour predictions of U-Net KP, we obtained the bias and 95% limits of agreement of 0.8 ± 10.8% for the right ventricular fractional area change measurements, -0.04 ± 0.54 cm for the tricuspid annular plane systolic excursion measurements and 0.2 ± 6.6% for the right ventricular free wall strain measurements. These results were also comparable to the semi-automatically derived inter-observer discrepancies of 0.4 ± 11.8%, -0.37 ± 0.58 cm and -1.0 ± 7.7% for the aforementioned metrics, respectively. CONCLUSION: Given the appropriate data, automated segmentation and quantification of the right ventricle in 2-D echocardiography are feasible with existing methods. However, keypoint detection architectures may offer higher robustness and information density for the same computational cost.

Mechanical Wave Velocities in Left Ventricular Walls in Healthy Subjects and Patients With Aortic Stenosis.

BACKGROUND: Mechanical wave velocity (MWV) measurement is a promising method for evaluating myocardial stiffness, because these velocities are higher in patients with myocardial disease. OBJECTIVES: Using high frame rate echocardiography and a novel method for detection of myocardial mechanical waves, this study aimed to estimate the MWVs for different left ventricular walls and events in healthy subjects and patients with aortic stenosis (AS). Feasibility and reproducibility were evaluated. METHODS: This study included 63 healthy subjects and 13 patients with severe AS. All participants underwent echocardiographic examination including 2-dimensional high frame rate recordings using a clinical scanner. Cardiac magnetic resonance was performed in 42 subjects. The authors estimated the MWVs at atrial kick and aortic valve closure in different left ventricular walls using the clutter filter wave imaging method. RESULTS: Mechanical wave imaging in healthy subjects demonstrated the highest feasibility for the atrial kick wave reaching >93% for all 4 examined left ventricular walls. The MWVs were higher for the inferolateral and anterolateral walls (2.2 and 2.6 m/s) compared with inferoseptal and anteroseptal walls (1.3 and 1.6 m/s) (P < 0.05) among healthy subjects. The septal MWVs at aortic valve closure were significantly higher for patients with severe AS than for healthy subjects. CONCLUSIONS: MWV estimation during atrial kick is feasible and demonstrates higher velocities in the lateral walls, compared with septal walls. The authors propose indicators for quality assessment of the mechanical wave slope as an aid for achieving consistent measurements. The discrimination between healthy subjects and patients with AS was best for the aortic valve closure mechanical waves. (Ultrasonic Markers for Myocardial Fibrosis and Prognosis in Aortic Stenosis; NCT03422770).

Automatic measurements of left ventricular volumes and ejection fraction by artificial intelligence: clinical validation in real time and large databases.

AIMS: Echocardiography is a cornerstone in cardiac imaging, and left ventricular (LV) ejection fraction (EF) is a key parameter for patient management. Recent advances in artificial intelligence (AI) have enabled fully automatic measurements of LV volumes and EF both during scanning and in stored recordings. The aim of this study was to evaluate the impact of implementing AI measurements on acquisition and processing time and test-retest reproducibility compared with standard clinical workflow, as well as to study the agreement with reference in large internal and external databases. METHODS AND RESULTS: Fully automatic measurements of LV volumes and EF by a novel AI software were compared with manual measurements in the following clinical scenarios: (i) in real time use during scanning of 50 consecutive patients, (ii) in 40 subjects with repeated echocardiographic examinations and manual measurements by 4 readers, and (iii) in large internal and external research databases of 1881 and 849 subjects, respectively. Real-time AI measurements significantly reduced the total acquisition and processing time by 77% (median 5.3 min, P < 0.001) compared with standard clinical workflow. Test-retest reproducibility of AI measurements was superior in inter-observer scenarios and non-inferior in intra-observer scenarios. AI measurements showed good agreement with reference measurements both in real time and in large research databases. CONCLUSION: The software reduced the time taken to perform and volumetrically analyse routine echocardiograms without a decrease in accuracy compared with experts.

Automated 2-D and 3-D Left Atrial Volume Measurements Using Deep Learning.

OBJECTIVE: Echocardiography, a critical tool for assessing left atrial (LA) volume, often relies on manual or semi-automated measurements. This study introduces a fully automated, real-time method for measuring LA volume in both 2-D and 3-D imaging, in the aim of offering accuracy comparable to that of expert assessments while saving time and reducing operator variability. METHODS: We developed an automated pipeline comprising a network to identify the end-systole (ES) time point and robust 2-D and 3-D U-Nets for segmentation. We employed data sets of 789 2-D images and 286 3-D recordings and explored various training regimes, including recurrent networks and pseudo-labeling, to estimate volume curves. RESULTS: Our baseline results revealed an average volume difference of 2.9 mL for 2-D and 7.8 mL for 3-D, respectively, compared with manual methods. The application of pseudo-labeling to all frames in the cine loop generally led to more robust volume curves and notably improved ES measurement in cases with limited data. CONCLUSION: Our results highlight the potential of automated LA volume estimation in clinical practice. The proposed prototype application, capable of processing real-time data from a clinical ultrasound scanner, provides valuable temporal volume curve information in the echo lab.

Echocardiographic Reference Ranges of Global Longitudinal Strain for All Cardiac Chambers Using Guideline-Directed Dedicated Views.

BACKGROUND: Myocardial deformation by echocardiographic strain imaging is a key measurement in cardiology, providing valuable diagnostic and prognostic information. Reference ranges for strain should be established from large healthy populations with minimal methodologic biases and variability. OBJECTIVES: The aim of this study was to establish echocardiographic reference ranges, including lower normal limits of global strains for all 4 cardiac chambers, by guideline-directed dedicated views from a large healthy population and to evaluate the influence of subject-specific characteristics on strain. METHODS: In total, 1,329 healthy participants from HUNT4Echo, the echocardiographic substudy of the 4th wave of the Trøndelag Health Study, were included. Echocardiographic recordings specific for each chamber were optimized according to current recommendations. Two experienced sonographers recorded all echocardiograms using GE HealthCare Vivid E95 scanners. Analyses were performed by experts using GE HealthCare EchoPAC. RESULTS: The reference ranges for left ventricular (LV) global longitudinal strain and right ventricular free-wall strain were -24% to -16% and -35% to -17%, respectively. Correspondingly, left atrial (LA) and right atrial (RA) reservoir strains were 17% to 49% and 17% to 59%. All strains showed lower absolute values with higher age, except for LA and RA contractile strains, which were higher. The feasibility for strain was overall good (LV 96%, right ventricular 83%, LA 94%, and RA 87%). All chamber-specific strains were associated with age, and LV strain was associated with sex. CONCLUSIONS: Reference ranges of strain for all cardiac chambers were established based on guideline-directed chamber-specific recordings. Age and sex were the most important factors influencing reference ranges and should be considered when using strain echocardiography.

Flow Dynamics in Children With Bicuspid Aortic Valve: A Blood Speckle Tracking Study.

OBJECTIVE: Bicuspid aortic valve (BAV) is associated with progressive aortic dilation. Although the etiology is complex, altered flow dynamics is thought to play an important role. Blood speckle tracking (BST) allows for visualization and quantification of complex flow, which could be useful in identifying patients at risk of root dilation and could aid in surgical planning. The aims of this study were to assess and quantify flow in the aortic root and left ventricle using BST in children with bicuspid aortic valves. METHODS AND RESULTS: A total of 38 children <10 y of age were included (24 controls, 14 with BAV). Flow dynamics were examined using BST in the aortic root and left ventricle. Children with BAV had altered systolic flow patterns in the aortic root and higher aortic root average vorticity (25.9 [23.4-29.2] Hz vs. 17.8 [9.0-26.2] Hz, p < 0.05), vector complexity (0.17 [0.14-0.31] vs. 0.05 [0.02-0.13], p < 0.01) and rate of energy loss (7.9 [4.9-12.1] mW/m vs. 2.7 [1.2-7.4] mW/m, p = 0.01). Left ventricular average diastolic vorticity (20.9 ± 5.8 Hz vs. 11.4 ± 5.2 Hz, p < 0.01), kinetic energy (0.11 ± 0.05 J/m vs. 0.04 ± 0.02 J/m, p < 0.01), vector complexity (0.38 ± 0.1 vs. 0.23 ± 0.1, p < 0.01) and rate of energy loss (11.1 ± 4.8 mW/m vs. 2.7 ± 1.9 mW/m, p < 0.01) were higher in children with BAV. CONCLUSION: Children with BAV exhibit altered flow dynamics in the aortic root and left ventricle in the absence of significant aortic root dilation. This may represent a substrate and potential predictor for future dilation and diastolic dysfunction.

Retrospective Ultrasound Doppler Quantification Using a Single Acquisition in Healthy Adults.

OBJECTIVE: Using an experimental tool for retrospective ultrasound Doppler quantification-with high temporal resolution and large spatial coverage-simultaneous flow and tissue measurements were obtained. We compared and validated these experimental values against conventional measurements to determine if the experimental acquisition produced trustworthy tissue and flow velocities. METHODS: We included 21 healthy volunteers. The only exclusion criterion was the presence of an irregular heartbeat. Two ultrasound examinations were performed for each participant, one using conventional and one using experimental acquisition. The experimental acquisition used multiple plane wave emissions combined with electrocardiography stitching to obtain continuous data with over 3500 frames per second. With two recordings covering a biplane apical view of the left ventricle, we retrospectively extracted selected flow and tissue velocities. RESULTS: Flow and tissue velocities were compared between the two acquisitions. Statistical testing showed a small but significant difference. We also exemplified the possibility of extracting spectral tissue Doppler from different sample volumes in the myocardium within the imaging sector, showing a decrease in the velocities from the base to the apex. CONCLUSION: This study demonstrates the feasibility of simultaneous, retrospective spectral and color Doppler of both tissue and flow from an experimental acquisition covering a full sector width. The measurements were significantly different between the two acquisitions but were still comparable, as the biases were small compared to clinical practice, and the two acquisitions were not done simultaneously. The experimental acquisition also enabled the study of deformation by simultaneous spectral velocity traces from all regions of the image sector.

Normalized Echocardiographic Values From Guideline-Directed Dedicated Views for Cardiac Dimensions and Left Ventricular Function.

BACKGROUND: Continuous technologic development and updated recommendations for image acquisitions creates a need to update the current normal reference ranges for echocardiography. The best method of indexing cardiac volumes is unknown. OBJECTIVES: The authors used 2- and 3-dimensional echocardiographic data from a large cohort of healthy individuals to provide updated normal reference data for dimensions and volumes of the cardiac chambers as well as central Doppler measurements. METHODS: In the fourth wave of the HUNT (Trøndelag Health) study in Norway 2,462 individuals underwent comprehensive echocardiography. Of these, 1,412 (55.8% women) were classified as normal and formed the basis for updated normal reference ranges. Volumetric measures were indexed to body surface area and height in powers of 1 to 3. RESULTS: Normal reference data for echocardiographic dimensions, volumes, and Doppler measurements were presented according to sex and age. Left ventricular ejection fraction had lower normal limits of 50.8% for women and 49.6% for men. According to sex-specific age groups, the upper normal limits for left atrial end-systolic volume indexed to body surface area ranged from 44 mL/m2 to 53 mL/m2, and the corresponding upper normal limit for right ventricular basal dimension ranged from 43 mm to 53 mm. Indexing to height raised to the power of 3 accounted for more of the variation between sexes than indexing to body surface area. CONCLUSIONS: The authors present updated normal reference values for a wide range of echocardiographic measures of both left- and right-side ventricular and atrial size and function from a large healthy population with a wide age-span. The higher upper normal limits for left atrial volume and right ventricular dimension highlight the importance of updating reference ranges accordingly following refinement of echocardiographic methods.

Deep Learning for Improved Precision and Reproducibility of Left Ventricular Strain in Echocardiography: A Test-Retest Study.

AIMS: Assessment of left ventricular (LV) function by echocardiography is hampered by modest test-retest reproducibility. A novel artificial intelligence (AI) method based on deep learning provides fully automated measurements of LV global longitudinal strain (GLS) and may improve the clinical utility of echocardiography by reducing user-related variability. The aim of this study was to assess within-patient test-retest reproducibility of LV GLS measured by the novel AI method in repeated echocardiograms recorded by different echocardiographers and to compare the results to manual measurements. METHODS: Two test-retest data sets (n = 40 and n = 32) were obtained at separate centers. Repeated recordings were acquired in immediate succession by 2 different echocardiographers at each center. For each data set, 4 readers measured GLS in both recordings using a semiautomatic method to construct test-retest interreader and intrareader scenarios. Agreement, mean absolute difference, and minimal detectable change (MDC) were compared to analyses by AI. In a subset of 10 patients, beat-to-beat variability in 3 cardiac cycles was assessed by 2 readers and AI. RESULTS: Test-retest variability was lower with AI compared with interreader scenarios (data set I: MDC = 3.7 vs 5.5, mean absolute difference = 1.4 vs 2.1, respectively; data set II: MDC = 3.9 vs 5.2, mean absolute difference = 1.6 vs 1.9, respectively; all P < .05). There was bias in GLS measurements in 13 of 24 test-retest interreader scenarios (largest bias, 3.2 strain units). In contrast, there was no bias in measurements by AI. Beat-to-beat MDCs were 1,5, 2.1, and 2.3 for AI and the 2 readers, respectively. Processing time for analyses of GLS by the AI method was 7.9 ± 2.8 seconds. CONCLUSION: A fast AI method for automated measurements of LV GLS reduced test-retest variability and removed bias between readers in both test-retest data sets. By improving the precision and reproducibility, AI may increase the clinical utility of echocardiography.

Clinical Influence of Handheld Ultrasound, Supported by Automatic Quantification and Telemedicine, in Suspected Heart Failure.

Early and correct heart failure (HF) diagnosis is essential to improvement of patient care. We aimed to evaluate the clinical influence of handheld ultrasound device (HUD) examinations by general practitioners (GPs) in patients with suspected HF with or without the use of automatic measurement of left ventricular (LV) ejection fraction (autoEF), mitral annular plane systolic excursion (autoMAPSE) and telemedical support. Five GPs with limited ultrasound experience examined 166 patients with suspected HF (median interquartile range = 70 (63-78) y; mean ± SD EF = 53 ± 10%). They first performed a clinical examination. Second, they added an examination with HUD, automatic quantification tools and, finally, telemedical support by an external cardiologist. At all stages, the GPs considered whether the patients had HF. The final diagnosis was made by one of five cardiologists using medical history and clinical evaluation including a standard echocardiography. Compared with the cardiologists' decision, the GPs correctly classified 54% by clinical evaluation. The proportion increased to 71% after adding HUDs, and to 74 % after telemedical evaluation. Net reclassification improvement was highest for HUD with telemedicine. There was no significant benefit of the automatic tools (p ≥ 0.58). Addition of HUD and telemedicine improved the GPs' diagnostic precision in suspected HF. Automatic LV quantification added no benefit. Refined algorithms and more training may be needed before inexperienced users benefit from automatic quantification of cardiac function by HUDs.

Assessment of Early Diastolic Intraventricular Pressure Difference in Children by Blood Speckle-Tracking Echocardiography.

BACKGROUND: The lack of reliable echocardiographic techniques to assess diastolic function in children is a major clinical limitation. Our aim was to develop and validate the intraventricular pressure difference (IVPD) calculation using blood speckle-tracking (BST) and investigate the method's potential role in the assessment of diastolic function in children. METHODS: Blood speckle-tracking allows two-dimensional angle-independent blood flow velocity estimation. Blood speckle-tracking images of left ventricular (LV) inflow from the apical 4-chamber view in 138 controls, 10 patients with dilated cardiomyopathies (DCMs), and 21 patients with hypertrophic cardiomyopathies (HCMs) <18 years of age were analyzed to study LV IVPD during early diastole. Reproducibility of the IVPD analysis was assessed, IVPD estimates from BST and color M mode were compared, and the validity of the BST-based IVPD calculations was tested in a computer flow model. RESULTS: Mean IVPD was significantly higher in controls (-2.28 ± 0.62 mm Hg) compared with in DCM (-1.21 ± 0.39 mm Hg, P < .001) and HCM (-1.57 ± 0.47 mm Hg, P < .001) patients. Feasibility was 88.3% in controls, 80% in DCM patients, and 90.4% in HCM patients. The peak relative negative pressure occurred earlier at the apex than at the base and preceded the peak E-wave LV filling velocity, indicating that it represents diastolic suction. Intraclass correlation coefficients for intra- and interobserver variability were 0.908 and 0.702, respectively. There was a nonsignificant mean difference of 0.15 mm Hg between IVPD from BST and color M mode. Estimation from two-dimensional velocities revealed a difference in peak IVPD of 0.12 mm Hg (6.6%) when simulated in a three-dimensional fluid mechanics model. CONCLUSIONS: Intraventricular pressure difference calculation from BST is highly feasible and provides information on diastolic suction and early filling in children with heart disease. Intraventricular pressure difference was significantly reduced in children with DCM and HCM compared with controls, indicating reduced early diastolic suction in these patient groups.

Real-Time Echocardiography Guidance for Optimized Apical Standard Views.

Measurements of cardiac function such as left ventricular ejection fraction and myocardial strain are typically based on 2-D ultrasound imaging. The reliability of these measurements depends on the correct pose of the transducer such that the 2-D imaging plane properly aligns with the heart for standard measurement views and is thus dependent on the operator's skills. We propose a deep learning tool that suggests transducer movements to help users navigate toward the required standard views while scanning. The tool can simplify echocardiography for less experienced users and improve image standardization for more experienced users. Training data were generated by slicing 3-D ultrasound volumes, which permits simulation of the movements of a 2-D transducer. Neural networks were further trained to calculate the transducer position in a regression fashion. The method was validated and tested on 2-D images from several data sets representative of a prospective clinical setting. The method proposed the adequate transducer movement 75% of the time when averaging over all degrees of freedom and 95% of the time when considering transducer rotation solely. Real-time application examples illustrate the direct relation between the transducer movements, the ultrasound image and the provided feedback.

Real-time guiding by deep learning during echocardiography to reduce left ventricular foreshortening and measurement variability.

Aims: Apical foreshortening leads to an underestimation of left ventricular (LV) volumes and an overestimation of LV ejection fraction and global longitudinal strain. Real-time guiding using deep learning (DL) during echocardiography to reduce foreshortening could improve standardization and reduce variability. We aimed to study the effect of real-time DL guiding during echocardiography on measures of LV foreshortening and inter-observer variability. Methods and results: Patients (n = 88) in sinus rhythm referred for echocardiography without indication for contrast were included. All participants underwent three echocardiograms. The first two examinations were performed by sonographers, and the third by cardiologists. In Period 1, the sonographers were instructed to provide high-quality echocardiograms. In Period 2, the DL guiding was used by the second sonographer. One blinded expert measured LV length in all recordings. Tri-plane recordings by cardiologists were used as reference. Apical foreshortening was calculated at the end-diastole. Both sonographer groups significantly foreshortened the LV in Period 1 (mean foreshortening: Sonographer 1: 4 mm; Sonographer 2: 3 mm, both P < 0.001 vs. reference) and reduced foreshortening in Period 2 (2 and 0 mm, respectively. Period 1 vs. Period 2, P < 0.05). Sonographers using DL guiding did not foreshorten more than cardiologists (P ≥ 0.409). Real-time guiding did not improve intra-class correlation (ICC) [LV end-diastolic volume ICC, (95% confidence interval): DL guiding 0.87 (0.77-0.93) vs. no guiding 0.92 (0.88-0.95)]. Conclusion: Real-time guiding reduced foreshortening among experienced operators and has the potential to improve image standardization. Even though the effect on inter-operator variability was minimal among experienced users, real-time guiding may improve test-retest variability among less experienced users. Clinical trial registration: ClinicalTrials.gov, Identifier: NCT04580095.

Real-time guidance by deep learning of experienced operators to improve the standardization of echocardiographic acquisitions.

Aims: Impaired standardization of echocardiograms may increase inter-operator variability. This study aimed to determine whether the real-time guidance of experienced sonographers by deep learning (DL) could improve the standardization of apical recordings. Methods and results: Patients (n = 88) in sinus rhythm referred for echocardiography were included. All participants underwent three examinations, whereof two were performed by sonographers and the third by cardiologists. In the first study period (Period 1), the sonographers were instructed to provide echocardiograms for the analyses of the left ventricular function. Subsequently, after brief training, the DL guidance was used in Period 2 by the sonographer performing the second examination. View standardization was quantified retrospectively by a human expert as the primary endpoint and the DL algorithm as the secondary endpoint. All recordings were scored in rotation and tilt both separately and combined and were categorized as standardized or non-standardized. Sonographers using DL guidance had more standardized acquisitions for the combination of rotation and tilt than sonographers without guidance in both periods (all P ≤ 0.05) when evaluated by the human expert and DL [except for the apical two-chamber (A2C) view by DL evaluation]. When rotation and tilt were analysed individually, A2C and apical long-axis rotation and A2C tilt were significantly improved, and the others were numerically improved when evaluated by the echocardiography expert. Furthermore, all, except for A2C rotation, were significantly improved when evaluated by DL (P < 0.01). Conclusion: Real-time guidance by DL improved the standardization of echocardiographic acquisitions by experienced sonographers. Future studies should evaluate the impact with respect to variability of measurements and when used by less-experienced operators. ClinicalTrialsgov Identifier: NCT04580095.