Automatic assessment of left ventricular function for hemodynamic monitoring using artificial intelligence and transesophageal echocardiography.

We have developed a method to automatically assess LV function by measuring mitral annular plane systolic excursion (MAPSE) using artificial intelligence and transesophageal echocardiography (autoMAPSE). Our aim was to evaluate autoMAPSE as an automatic tool for rapid and quantitative assessment of LV function in critical care patients. In this retrospective study, we studied 40 critical care patients immediately after cardiac surgery. First, we recorded a set of echocardiographic data, consisting of three consecutive beats of midesophageal two- and four-chamber views. We then altered the patient's hemodynamics by positioning them in anti-Trendelenburg and repeated the recordings. We measured MAPSE manually and used autoMAPSE in all available heartbeats and in four LV walls. To assess the agreement with manual measurements, we used a modified Bland-Altman analysis. To assess the precision of each method, we calculated the least significant change (LSC). Finally, to assess trending ability, we calculated the concordance rates using a four-quadrant plot. We found that autoMAPSE measured MAPSE in almost every set of two- and four-chamber views (feasibility 95%). It took less than a second to measure and average MAPSE over three heartbeats. AutoMAPSE had a low bias (0.4 mm) and acceptable limits of agreement (- 3.7 to 4.5 mm). AutoMAPSE was more precise than manual measurements if it averaged more heartbeats. AutoMAPSE had acceptable trending ability (concordance rate 81%) during hemodynamic alterations. In conclusion, autoMAPSE is feasible as an automatic tool for rapid and quantitative assessment of LV function, indicating its potential for hemodynamic monitoring.

Fully digital self-screening for atrial fibrillation with patch electrocardiogram.

AIMS: Atrial fibrillation (AF) is the most common arrhythmia worldwide. The AF is associated with severe mortality, morbidity, and healthcare costs, and guidelines recommend screening people at risk. However, screening methods and organization still need to be clarified. The current study aimed to assess the feasibility of a fully digital self-screening procedure and to assess the prevalence of undetected AF using a continuous patch electrocardiogram (ECG) monitoring system. METHODS AND RESULTS: Individuals ≥65 years old with at least one additional risk factor for stroke from the general population of Norway were invited to a fully digital continuous self-screening for AF using a patch ECG device (ECG247 Smart Heart Sensor). Participants self-reported clinical characteristics and usability online, and all participants received digital feedback of their results. A total of 2118 individuals with a mean CHA2DS2-VASc risk score of 2.6 (0.9) were enrolled in the study [74% women; mean age 70.1 years (4.2)]. Of these, 1849 (87.3%) participants completed the ECG self-screening test, while 215 (10.2%) did not try to start the test and 54 (2.5%) failed to start the test. The system usability score was 84.5. The mean ECG monitoring time was 153 h (87). Atrial fibrillation was detected in 41 (2.2%) individuals. CONCLUSION: This fully digitalized self-screening procedure for AF demonstrated excellent feasibility. The number needed to screen was 45 to detect one unrecognized case of AF in subjects at risk for stroke. Randomized studies with long-term follow-up are needed to assess whether self-screening for AF can reduce the incidence of AF-related complications. CLINICAL TRIALS: NCT04700865.

Treadmill running intensity and post-exercise increase in plasma cardiac troponin I and T-A pilot study in healthy volunteers.

BACKGROUND: Plasma concentrations of cardiac troponins increase in healthy individuals after strenuous training, but the response to lower exercise intensities has not been characterized. AIM: To determine whether exercise at moderate intensity significantly increases plasma cardiac troponins measured with different assays in healthy recreational athletes. METHODS: Twenty-four self-reported healthy volunteers were instructed to complete three 60-min bouts of treadmill running at variable intensities: High-intensity training (HIT) including a maximal exercise test and an anaerobic threshold test followed by training at 80%-95% of maximum heart rate (HRmax ), Moderate-intensity training (MIT) at 60%-75% of HRmax , and Low-intensity training (LIT) at 45%-55% of HRmax . Blood samples were collected before and at 2, 4, and 6 h after HIT and 4 h after MIT and LIT. Troponin I and T were measured in plasma samples with assays from Abbot, Siemens, and Roche. RESULTS: Plasma troponins measured with all assays were significantly increased compared to baseline after HIT but not after LIT. After HIT, the fraction of all participants with one or more values above the assay-specific 99th percentiles ranged from 13% to 61%. The biomarker criteria for acute myocardial injury were met after HIT for troponin T in 75% of female participants having no clinical evidence of coronary artery disease. CONCLUSION: High-intensity, but not moderate- or low-intensity, training for 60 min induced a potentially clinically significant increase in plasma cardiac troponins in healthy volunteers. Results exceeding the population 99th percentiles were most frequent with the troponin T assay.

Increased deformation of the left ventricle during exercise test measured by global longitudinal strain can rule out significant coronary artery disease in patients with suspected unstable angina pectoris.

BACKGROUND: Noninvasive identification of significant coronary artery disease (CAD) in patients with unstable angina pectoris (UAP) is challenging. Exercise stress testing has been used for years in patients with suspected CAD but has low diagnostic accuracy. The use of Global longitudinal strain (GLS) by speckle tracking echocardiography is a highly sensitive and reproducible parameter for detection of myocardial ischemia. Our aim was to study if identification of normal or ischemic myocardium by measurement of GLS immediately after an ordinary bicycle exercise stress testing in patients with suspected UAP could identify or rule out significant CAD. METHODS: Seventy-eight patients referred for coronary angiography from outpatient clinics and the emergency department with chest pain, inconclusive ECG and normal values of Troponin-T was included. All patients underwent echocardiographic examination at rest and immediately after maximum stress by exercise on a stationary bicycle. Significant CAD was defined by diameter stenosis > 90% by coronary angiography. In patients with coronary stenosis between 50-90%, fractional flow reserve (FFR) was measured and defined abnormal < .80. Analysis of echocardiographic data were performed blinded for angiographic data. Patients were discharged diagnosed with CAD (n = 34) or non-coronary chest pain (NCCP, n = 44). RESULTS: In patients with NCCP, GLS at rest was -21.1 ± 1.7% and -25.5 ± 2.6% at maximum stress (P < .01). In patients with CAD, GLS at rest was -16.8 ± 4.0% and remained unchanged at maximum stress (-16.6 ± 4.6%, P = .69). In patients with NCCP, LVEF was 56.1% ± 6.0 and increased to 61.8% 5.2, P < .01. In CAD patients, LVEF at rest was 54.7% ± 8.6 and increased to 58.2% ± 9.5 during stress, P = .16. In NCCP patients, Wall Motion Score index decreased .02 ± .07, P = .03 during stress and was without significant changes in patients with CAD. Area under the curve (AUC) for distinguishing CAD for was .97 (.95-1.00), .63 (.49-.76), and .71 (.59-.83) for GLS, LVEF, and WMSi, respectively. CONCLUSION: In patients with suspected UAP, increased deformation of the left ventricle measured by GLS immediately after exercise stress testing identified normal myocardium without CAD. Reduced LV contractile function by GLS without increase after exercise identified significant CAD.

Feasibility of prehospital identification of non-ST-elevation myocardial infarction by ECG, troponin and echocardiography.

BACKGROUND: Prehospital identification and selection of treatment strategy in patients with suspected non-ST-elevation myocardial infarction (NSTEMI) is challenging. The aim of this study was to evaluate the feasibility and diagnostic accuracy of prehospital ECG, troponin T (TnT) and transthoracic echocardiography (TTE) acquired by paramedics in early identification of NSTEMI. METHODS: Consecutive patients requesting an ambulance from Sorlandet Hospital, Norway due to chest pain between November 2017 and January 2020 were screened for inclusion in the study. One ambulance was equipped with ECG recorder, point-of-care TnT test and TTE scanner, and six paramedics were given necessary training. ECG, TnT result and TTE images were acquired prehospitally and transferred to an in-hospital cardiologist. NSTEMI was suspected in patients with ischaemic ECG changes, elevated TnT or myocardial regional wall motion abnormalities (RWMA) at TTE. RESULTS: A total of 253 patients were included in the study. ECG was interpretable by cardiologists in 243 (96%), TnT in 238 (94%) and TTE images in 240 (95%) patients. NSTEMI was the discharge diagnosis in 22 (9%) of these patients. Four (18%) patients with NSTEMI had ischaemic ECG changes, elevated TnT and RWMA at TTE. Eight (36%) patients with NSTEMI had positive findings at two of the diagnostic methods, six (27%) patients had positive findings at one, and four (18%) patients had no positive findings in any of the diagnostic methods. In three patients (14%) with NSTEMI, RWMA was the only positive test. The negative and positive predictive values for RWMA were 42% and 96%, respectively. CONCLUSIONS: Prehospital acquisition of ECG, TnT and interpretable TTE images by paramedics were feasible in most patients with chest pain. Based on these examinations, it was possible to identify the majority of cases with NSTEMI prehospitally and admit the patients directly to a hospital with facilities for percutaneous coronary intervention (PCI) for further treatment. TRIAL REGISTRATION NUMBER: NCT04223986.

Observational study on chest pain during the Covid-19 pandemic: changes and characteristics of visits to a Norwegian emergency department during the lockdown.

BACKGROUND: Following the spread of the Covid-19 pandemic in 2020, reports emerged on decreasing emergency department (ED) visits in many countries. Patients experiencing chest pain was no exception. The aim of the current study was to describe how the Covid-19 pandemic and the subsequential lockdown impacted the chest pain population in a Norwegian ED. METHODS: All patients presenting to the ED with chest pain during the study period were included. Data were collected retrospectively from the time period January 6th to August 30th, 2020, and compared to the corresponding period in 2019, assessing variations in the number of ED visits, severity, gender, and age. RESULTS: Fewer patients with chest pain were seen in the ED following the national lockdown in Norway, compared to the corresponding 2019 period (week 13: 38% fewer; weeks 11-27: 16% fewer). By week 28, the rate normalized compared to 2019 levels. There was a relative increase in lower acuity patients among these patients, while fewer moderate acuity patients were seen. During the initial period following lockdown, the median age was lower compared to the corresponding 2019 period (58 years (IQR 25) vs 62 years (IQR 24), respectively). Admissions due to acute coronary syndromes (ACS) remained proportionally stable. CONCLUSIONS: Succeeding the Covid-19 outbreak and the subsequent national lockdown in Norway, fewer chest pain patients presented to the ED. Paradoxically, the patients seemed to be less severely ill and were on average younger compared to 2019 data. However, the proportion of patients admitted with ACS was stable during this period. This could imply that some patients may have failed to seek medical advice despite experiencing a myocardial infarction.

Establishing the 99th percentile of a novel assay for high-sensitivity troponin I in a healthy blood donor population.

Background: The recommended cut-off of cardiac troponin (cTn) for the diagnosis of acute myocardial infarction (AMI) is the 99th percentile in a healthy reference population. We aimed to determine the 99th percentile of the novel ADVIA Centaur® High Sensitivity Troponin I assay (Siemens Healthcare Diagnostics) in fresh lithium heparin plasma samples from healthy blood donors. Methods: A total of 1000 apparently healthy blood donors were included. High-sensitivity (hs) cTnI, hs-cTnT, creatinine and N-terminal pro b-type natriuretic peptide (NT-proBNP) were measured in fresh lithium heparin plasma samples, and glycated hemoglobin (HbA1c) was measured in ethylenediaminetetraacetic acid (EDTA)-blood. The 99th percentile was estimated for the whole population, as well as for males and females separately. Results: For the total population the 99th percentile of ADVIA Centaur® High Sensitivity Troponin I was 96 (65-149) ng/L. The estimated value differed significantly from results published by others and was highly dependent on which values were considered statistical outliers. Conclusions: The estimated 99th percentile for hs-cTnI in the population studied differed significantly from previously published results. There is a need for further specifications regarding how subjects used for estimating the 99th percentile of cTns in healthy populations should be recruited and how outlier values should be identified, as this can highly influence the diagnostic cut-off applied for AMI.

Completeness and correctness of acute myocardial infarction diagnoses in a medical quality register and an administrative health register.

Aims: Health registers are used for administrative purposes, disease surveillance, quality assessment, and research. The value of the registers is entirely dependent on the quality of their data. The aim of this study was to investigate and compare the completeness and correctness of the acute myocardial infarction (AMI) diagnosis in the Norwegian Myocardial Infarction Register and in the Norwegian Patient Register. Methods: All Norwegian patients admitted directly to St Olavs hospital, Trondheim University Hospital, Trondheim University Hospital from 1 July to 31 December 2012 and who had plasma levels of cardiac troponin T measured during their hospitalization (n=4835 unique individuals, n=5882 hospitalizations) were identified in the hospital biochemical database. A gold standard for AMI was established by evaluation of maximum troponin T levels and by review of the information in the medical records. Cases of AMI in the registers were classified as true positive, false positive, true negative, and false negative according to the gold standard. We calculated sensitivity, positive predictive value (PPV), specificity, and negative predictive value (NPV). Results: The Norwegian Myocardial Infarction Register had a sensitivity of 86.0% (95% confidence interval (CI) 82.8-89.3%), PPV of 97.9% (96.4-99.3%), and specificity of 99.9% and NPV of 98.9% (98.6-99.2%) (99.8-100%). The corresponding figures for the Norwegian Patient Register were 85.8% (95% CI 82.5-89.1%), 95.1% (92.9-97.2%), and 99.7% (99.5-99.8%) and 98.9% (98.6-99.2%), respectively. Both registers had a sensitivity higher than 95% when compared to hospital discharge diagnoses. The results were similar for men and women and for cases below and above 80 years of age. Conclusions: The Norwegian Myocardial Infarction Register and the Norwegian Patient Register are adequately complete and correct for administrative purposes, disease surveillance, quality assessment, and research.

Description of chest pain patients in a Norwegian emergency department.

OBJECTIVES: Chest pain is one of the most frequent chief complaints in patients admitted to the emergency department (ED). No study has evaluated this patient population in Norway. The aim of the present study was to describe the diagnoses and logistics of chest pain patients admitted to the ED at a Norwegian university hospital. DESIGN: Data were collected from all patients referred to the ED at St. Olav's Hospital-Trondheim University Hospital in 2015. Patients presenting at the ED with a chief complaint of chest pain based on the Rapid Emergency Triage and Treatment System were included in this study. RESULTS: In the study period, 22,403 patients were admitted to the ED. Chest pain was the second most common chief complaint (11%), only exceeded by acute abdominal pain (18%). Among the chest pain patients, 12% had acute coronary syndrome (ACS) and 37% were diagnosed with non-specific chest pain. Gastrointestinal, musculoskeletal, and pulmonary conditions were less common, affecting 6%, 4%, and 4% of the patients, respectively. In total, 90% of the ED chest pain patients were handled by the Clinic of Cardiology. CONCLUSIONS: Chest pain was the second most common chief complaint among patients admitted to the ED, and most patients were handled by the Clinic of Cardiology. More than one-third of the patients received a cardiac diagnosis, but the largest group of patients was discharged with a diagnosis of non-specific chest pain. Other conditions managed by non-cardiac medical specialties were less common.

Global longitudinal strain is a more reproducible measure of left ventricular function than ejection fraction regardless of echocardiographic training.

BACKGROUND: Left ventricular ejection fraction (LVEF) is an established method for evaluation of left ventricular (LV) systolic function. Global longitudinal strain (GLS) by speckle tracking echocardiography seems to be an important additive method for evaluation of LV function with improved reproducibility compared with LVEF. Our aim was to compare reproducibility of GLS and LVEF between an expert and trainee both as echocardiographic examiner and analyst. METHODS: Forty-seven patients with recent Acute Coronary Syndrome (ACS) underwent echocardiographic examination by both an expert echocardiographer and a trainee. Both echocardiographers, blinded for clinical data and each other's findings, performed image analysis for evaluation of intra- and inter- observer variability. GLS was measured using speckle tracking echocardiography. LVEF was calculated by Simpson's biplane method. RESULTS: The trainee measured a GLS of - 19.4% (±3.5%) and expert - 18.7% (±3.2%) with an Intra class correlation coefficient (ICC) of 0.89 (0.74-0.95). LVEF by trainee was 50.3% (±8.2%) and by expert 53.6% (±8.6%), ICC coefficient was 0.63 (0.32-0.80). For GLS the systematic difference was 0.21% (- 4.58-2.64) vs. 4.08% (- 20.78-12.62) for LVEF. CONCLUSION: GLS is a more reproducible method for evaluation of LV function than LVEF regardless of echocardiographic training.

Are the guidelines for treatment of myocardial infarction complied with?

BACKGROUND: New guidelines recommend early invasive evaluation and treatment for most patients with acute myocardial infarction--including patients with myocardial infarction without ST elevation in the ECG. This study examines compliance with the new guidelines at Sørlandet Hospital Arendal. MATERIAL AND METHOD: All patients admitted to Sørlandet Hospital Arendal with acute myocardial infarction in 2012 were registered in the Norwegian Myocardial Infarction Register. Data from the register were used to analyse the time that passed from symptom onset to coronary angiography and revascularisation. RESULTS: In 2012, 788 patients were admitted to Sørlandet Hospital Arendal with acute myocardial infarction. Of these, 269 (34.1%) had ST elevation mycardial infarction (STEMI) and 519 (65.9%) had non-ST elevation myocardial infarction (NSTEMI). Most patients with ST elevation infarction (220 (81.8%)) were admitted directly to Sørlandet Hospital Arendal, and the median time from admission to revascularisation was 31 minutes. 347 (66.9%) of the patients with non-ST elevation infarction were first admitted to a local hospital before being transferred to Sørlandet Hospital Arendal. Only four (1.2%) of them underwent angiography within two hours of admission to the first hospital. 13 (9.0%) of the patients with non-ST elevation infarction who were admitted directly and underwent angiography (n = 144) had an angiogram within two hours of admission. Angiography was performed within 24 hours in 119 (34.3%) of those transferred (n = 347) and in 82 (56.9%) of the directly admitted patients who underwent angiography (n = 144). INTERPRETATION: Many patients with non-ST elevation infarction did not receive revascularisation with percutaneous coronary intervention (PCI) within the recommended time frame. Where there is a strong clinical suspicion of acute myocardial infarction, more patients should be admitted directly to hospitals with PCI preparedness.

Screening for Atrial Fibrillation by Digital Health Technology in Older People in Homecare Settings: A Feasibility Trial.

Aims: Users of homecare services are often excluded from clinical trials due to advanced age, multimorbidity, and frailty. Atrial fibrillation (AF) is a common and frequently undiagnosed arrhythmia in the elderly and is associated with severe mortality, morbidity, and healthcare costs. Timely identification prevents associated complications through evidence-based treatment. This study is aimed at assessing the feasibility of AF screening using new digital health technology in older people in a homecare setting. Methods: Users of homecare services ≥ 65 years old with at least one additional risk factor for stroke in two Norwegian municipalities were assessed for study participation by nurses. Participants performed a continuous prolonged ECG recording using a patch ECG device (ECG247 Smart Heart Sensor). Results: A total of 144 individuals were assessed for study participation, but only 18 (13%) were included. The main reasons for noninclusion were known AF and/or anticoagulation therapy (25%), severe cognitive impairment (26%), and lack of willingness to participate (36%). The mean age of participants performing the ECG test was 81 (SD ± 7) years, and 9 (50%) were women. All ECG tests were interpretable; the mean ECG monitoring time was 104 hours (IQR 34-338 hours). AF was detected in one individual (6%). Conclusion: This feasibility study highlights the challenges of enrolling older people receiving homecare services in clinical trials. However, all included participants performed an interpretable and prolonged continuous ECG recording with a digital ECG patch device. This trial is registered with NCT04700865.

Digital recruitment and compliance to treatment recommendations in the Norwegian Atrial Fibrillation self-screening pilot study.

Aims: Atrial fibrillation (AF) is prevalent, undiagnosed in approximately one-third of cases, and is associated with severe complications. Guidelines recommend screening individuals at increased risk of stroke. This report evaluated the digital recruitment procedure and compliance with the follow-up recommendations in participants with screen-detected AF in the Norwegian Atrial Fibrillation self-screening pilot study. Methods and results: Norwegians ≥65 years were invited through Facebooks posts, web pages, and newspapers to participate in the study. Targeted Facebook posts promoted over 11 days reached 84 208 users and 10 582 visitors to the study homepage. This accounted for 51% of the total homepage visitors (n = 20 704). A total of 2118 (10%) of the homepage visitors provided digital consent to participate after they met the inclusion criteria. The mean (standard deviation) age of the participants was 70 (4) years, and the majority [n = 1569 (74%)] were women. A total of 1849 (87%) participants completed the electrocardiogram self-screening test, identifying AF in 41 (2.2%) individuals. Of these, 39 (95%) participants consulted a general practitioner, and 34 (83%) participants initiated anticoagulation therapy. Conclusion: Digital recruitment and inclusion in digital AF screening with a high rate of initiation of anticoagulation therapy in AF positive screening cases are feasible. However, digital recruitment and inclusion may introduce selection bias with regard to age and gender. Larger studies are needed to determine the efficacy and cost-effectiveness of a fully digital AF screening. Trial registration: Clinical trials: NCT04700865.

Continuous monitoring of left ventricular function in postoperative intensive care patients using artificial intelligence and transesophageal echocardiography.

BACKGROUND: Continuous monitoring of mitral annular plane systolic excursion (MAPSE) using transesophageal echocardiography (TEE) may improve the evaluation of left ventricular (LV) function in postoperative intensive care patients. We aimed to assess the utility of continuous monitoring of LV function using TEE and artificial intelligence (autoMAPSE) in postoperative intensive care patients. METHODS: In this prospective observational study, we monitored 50 postoperative intensive care patients for 120 min immediately after cardiac surgery. We recorded a set of two-chamber and four-chamber TEE images every five minutes. We defined monitoring feasibility as how often the same wall from the same patient could be reassessed, and categorized monitoring feasibility as excellent if the same LV wall could be reassessed in ≥ 90% of the total recordings. To compare autoMAPSE with manual measurements, we rapidly recorded three sets of repeated images to assess precision (least significant change), bias, and limits of agreement (LOA). To assess the ability to identify changes (trending ability), we compared changes in autoMAPSE with the changes in manual measurements in images obtained during the initiation of cardiopulmonary bypass as well as before and after surgery. RESULTS: Monitoring feasibility was excellent in most patients (88%). Compared with manual measurements, autoMAPSE was more precise (least significant change 2.2 vs 3.1 mm, P < 0.001), had low bias (0.4 mm), and acceptable agreement (LOA - 2.7 to 3.5 mm). AutoMAPSE had excellent trending ability, as its measurements changed in the same direction as manual measurements (concordance rate 96%). CONCLUSION: Continuous monitoring of LV function was feasible using autoMAPSE. Compared with manual measurements, autoMAPSE had excellent trending ability, low bias, acceptable agreement, and was more precise.

Automatic Detection and Tracking of Anatomical Landmarks in Transesophageal Echocardiography for Quantification of Left Ventricular Function.

OBJECTIVE: Evaluation of left ventricular (LV) function in critical care patients is useful for guidance of therapy and early detection of LV dysfunction, but the tools currently available are too time-consuming. To resolve this issue, we previously proposed a method for the continuous and automatic quantification of global LV function in critical care patients based on the detection and tracking of anatomical landmarks on transesophageal heart ultrasound. In the present study, our aim was to improve the performance of mitral annulus detection in transesophageal echocardiography (TEE). METHODS: We investigated several state-of-the-art networks for both the detection and tracking of the mitral annulus in TEE. We integrated the networks into a pipeline for automatic assessment of LV function through estimation of the mitral annular plane systolic excursion (MAPSE), called autoMAPSE. TEE recordings from a total of 245 patients were collected from St. Olav's University Hospital and used to train and test the respective networks. We evaluated the agreement between autoMAPSE estimates and manual references annotated by expert echocardiographers in 30 Echolab patients and 50 critical care patients. Furthermore, we proposed a prototype of autoMAPSE for clinical integration and tested it in critical care patients in the intensive care unit. RESULTS: Compared with manual references, we achieved a mean difference of 0.8 (95% limits of agreement: -2.9 to 4.7) mm in Echolab patients, with a feasibility of 85.7%. In critical care patients, we reached a mean difference of 0.6 (95% limits of agreement: -2.3 to 3.5) mm and a feasibility of 88.1%. The clinical prototype of autoMAPSE achieved real-time performance. CONCLUSION: Automatic quantification of LV function had high feasibility in clinical settings. The agreement with manual references was comparable to inter-observer variability of clinical experts.

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.

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.

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.