Deep Learning-Derived Myocardial Strain.

BACKGROUND: Echocardiographic strain measurements require extensive operator experience and have significant intervendor variability. Creating an automated, open-source, vendor-agnostic method to retrospectively measure global longitudinal strain (GLS) from standard echocardiography B-mode images would greatly improve post hoc research applications and may streamline patient analyses. OBJECTIVES: This study was seeking to develop an automated deep learning strain (DLS) analysis pipeline and validate its performance across multiple applications and populations. METHODS: Interobserver/-vendor variation of traditional GLS, and simulated effects of variation in contour on speckle-tracking measurements were assessed. The DLS pipeline was designed to take semantic segmentation results from EchoNet-Dynamic and derive longitudinal strain by calculating change in the length of the left ventricular endocardial contour. DLS was evaluated for agreement with GLS on a large external dataset and applied across a range of conditions that result in cardiac hypertrophy. RESULTS: In patients scanned by 2 sonographers using 2 vendors, GLS had an intraclass correlation of 0.29 (95% CI: -0.01 to 0.53, P = 0.03) between vendor measurements and 0.63 (95% CI: 0.48-0.74, P < 0.001) between sonographers. With minor changes in initial input contour, step-wise pixel shifts resulted in a mean absolute error of 3.48% and proportional strain difference of 13.52% by a 6-pixel shift. In external validation, DLS maintained moderate agreement with 2-dimensional GLS (intraclass correlation coefficient [ICC]: 0.56, P = 0.002) with a bias of -3.31% (limits of agreement: -11.65% to 5.02%). The DLS method showed differences (P < 0.0001) between populations with cardiac hypertrophy and had moderate agreement in a patient population of advanced cardiac amyloidosis: ICC was 0.64 (95% CI: 0.53-0.72), P < 0.001, with a bias of 0.57%, limits of agreement of -4.87% to 6.01% vs 2-dimensional GLS. CONCLUSIONS: The open-source DLS provides lower variation than human measurements and similar quantitative results. The method is rapid, consistent, vendor-agnostic, publicly released, and applicable across a wide range of imaging qualities.

Right Ventricular Dysfunction Patterns among Patients with COVID-19 in the Intensive Care Unit: A Retrospective Cohort Analysis.

Rationale: Right ventricular (RV) dysfunction is common among patients hospitalized with coronavirus disease (COVID-19); however, its epidemiology may depend on the echocardiographic parameters used to define it. Objectives: To evaluate the prevalence of abnormalities in three common echocardiographic parameters of RV function among patients with COVID-19 admitted to the intensive care unit (ICU), as well as the effect of RV dilatation on differential parameter abnormality and the association of RV dysfunction with 60-day mortality. Methods: We conducted a retrospective cohort study of ICU patients with COVID-19 between March 4, 2020, and March 4, 2021, who received a transthoracic echocardiogram within 48 hours before to at most 7 days after ICU admission. RV dysfunction and dilatation, respectively, were defined by guideline thresholds for tricuspid annular plane systolic excursion (TAPSE), RV fractional area change, RV free wall longitudinal strain (RVFWS), and RV basal dimension or RV end-diastolic area. Association of RV dysfunction with 60-day mortality was assessed through logistic regression adjusting for age, prior history of congestive heart failure, invasive ventilation at the time of transthoracic echocardiogram, and Acute Physiology and Chronic Health Evaluation II score. Results: A total of 116 patients were included, of whom 69% had RV dysfunction by one or more parameters, and 36.3% of these had RV dilatation. The three most common patterns of RV dysfunction were the presence of three abnormalities, the combination of abnormal RVFWS and TAPSE, and isolated TAPSE abnormality. Patients with RV dilatation had worse RV fractional area change (24% vs. 36%; P = 0.001), worse RVFWS (16.3% vs. 19.1%; P = 0.005), higher RV systolic pressure (45 mm Hg vs. 31 mm Hg; P = 0.001) but similar TAPSE (13 mm vs. 13 mm; P = 0.30) compared with those with normal RV size. After multivariable adjustment, 60-day mortality was significantly associated with RV dysfunction (odds ratio, 2.91; 95% confidence interval, 1.01-9.44), as was the presence of at least two parameter abnormalities. Conclusions: ICU patients with COVID-19 had significant heterogeneity in RV function abnormalities present with different patterns associated with RV dilatation. RV dysfunction by any parameter was associated with increased mortality. Therefore, a multiparameter evaluation may be critical in recognizing RV dysfunction in COVID-19.

Novel left ventricular mechanical index in pulmonary arterial hypertension.

Ventricular interdependence plays an important role in pulmonary arterial hypertension (PAH). It can decrease left ventricular (LV) longitudinal strain (LVLS) and lead to a leftward displacement ("transverse shortening") of the interventricular septum (sTS). For this study, we hypothesized the ratio of LVLS/sTS would be a sensitive marker of systolic ventricular interactions in PAH. In a cross-sectional cohort of patients with PAH (n = 57) and matched controls (n = 57), we quantified LVLS and septal TS in the amplitude and time domain. We then characterized LV phenotypes using upset plots, ventricular interactions using network analysis, and longitudinal analysis in a representative cohort of 45 patients. We also measured LV metrics in mice subjected to pulmonary arterial banding (PAB) using a 7 T magnetic resonance imaging at baseline, Week 1, and Week 7 post-PAB (N = 9). Patients with PAH had significantly reduced absolute LVLS (15.4 ± 3.4 vs. 20.1 ± 2.3%, p < 0.0001), higher sTS (53.0 ± 12.2 vs. 28.0 ± 6.2%, p < 0.0001) and lower LVLS/sTS (0.30 ± 0.09 vs. 0.75 ± 0.16, p < 0.0001) compared to controls. Reduced LVLS/sTS was observed in 89.5% of patients, while diastolic dysfunction, impaired LVLS (<16%), and LV atrophy were observed in 73.7%, 52.6%, and 15.8%, respectively. In the longitudinal cohort, changes in LVLS/sTS were closely associated with changes in N-terminal pro B-type natriuretic peptide (r = 0.73, p < 0.0001) as well as survival. Mice subjected to PAB showed significant RV systolic dysfunction and decreased LVLS/sTS compared to sham animals. We conclude that in PAH, LVLV/sTS is a simple ratio that can reflect ventricular systolic interactions.

Reference change value of global longitudinal strain in clinical practice: A test-rest quality implementation project.

BACKGROUND: Reference change value (RCV) is used to assess the significance of the difference between two measurements after accounting for pre-analytic, analytic, and within-subject variability. The objective of the current study was to define the RCV for global longitudinal strain (GLS) using different semi-automated software in standard clinical practice. METHODS: Using a test-retest study design, we quantified the median coefficient of variation (CV) for GLS using AutoStrain and Automated Cardiac Motion Quantification (aCMQ) by Philips. Triplane left-ventricular ejection fraction (LVEF) was measured for comparison. Multivariable regression analysis was performed to determine factors influencing test-retest CV including image quality and the presence of segmental wall motion abnormalities (WMA). RCV was reported using a standard formula assuming two standard deviations for repeated measurements; results were also translated into Bayesian probability. Total measurement variation was described in terms of its three different components: pre-analytic (acquisition), analytic (measuring variation), and within-subject (biological) variation. RESULT: Of the 44 individuals who were screened, 41 had adequate quality for strain quantification. The mean age of the cohort was 56.4 ± 16.8 years, 41% female, LVEF was 55.8 ± 9.8% and the median and interquartile range for LV GLS was -17.2 [-19.3 to -14.8]%. Autostrain was more time efficient (80% less analysis time) and had a lower total median CV than aCMQ (CV = 7.4% vs. 17.6%, p < .001). The total CV was higher in patients with WMA (6.4% vs. 13.2%, p = .035). In non-segmental disease, the CV translates to a RCV of 15% (corresponding to a probability of real change of 80%). Assuming a within-subject variability of 4.0%, the component analysis identified that inter-reader variability accounts for 3.7% of the CV, while acquisition variability accounts for 4.0%. CONCLUSION: Using test-retest analysis and CVs, we find that an RCV of 15% for GLS represents an optimistic estimate in routine clinical practice. Based on our results, a higher RCV of 17%-21% is needed in order to provide a high probability of clinically meaningful change in GLS in all comers. The methodology presented here for determining measurement reproducibility and RCVs is easily translatable into clinical practice for any imaging parameter.

Optimizing right ventricular focused four-chamber views using three-dimensional imaging, a comparative magnetic resonance based study.

Obtaining focused right ventricular (RV) apical view remains challenging using conventional two-dimensional (2D) echocardiography. This study main objective was to determine whether measurements from RV focused views derived from three-dimensional (3D) echocardiography (3D-RV-focused) are closely related to measurements from magnetic resonance (CMR). A first cohort of 47 patients underwent 3D echocardiography and CMR imaging within 2 h of each other. A second cohort of 25 patients had repeat 3D echocardiography to determine the test-retest characteristics; and evaluate the bias associated with unfocused RV views. Tomographic views were extracted from the 3D dataset: RV focused views were obtained using the maximal RV diameter in the transverse plane, and unfocused views from a smaller transverse diameter enabling visualization of the tricuspid valve opening. Measures derived using the 3D-RV-focused view were strongly associated with CMR measurements. Among functional metrics, the strongest association was between RV fractional area change (RVFAC) and ejection fraction (RVEF) (r = 0.92) while tricuspid annular plane systolic excursion moderately correlated with RVEF (r = 0.47), all p < 0.001. Among RV size measures, the strongest association was found between RV end-systolic area (RVESA) and volume (r = 0.87, p < 0.001). RV unfocused views led on average to 10% underestimation of RVESA. The 3D-RV-focused method had acceptable test-retest characteristics with a coefficient of variation of 10% for RVESA and 11% for RVFAC. Deriving standardized RV focused views using 3D echocardiography strongly relates to CMR-derived measures and may improve reproducibility in RV 2D measurements.

CRITICAL CARE ECHO ROUNDS: Haemodynamic instability.

UNLABELLED: The use of echocardiography, whilst well established in cardiology, is a relatively new concept in critical care medicine. However, in recent years echocardiography's potential as both a diagnostic tool and a form of advanced monitoring in the critically ill patient has been increasingly recognised. In this series of Critical Care Echo Rounds, we explore the role of echocardiography in critical illness, beginning here with haemodynamic instability. We discuss the pathophysiology of the shock state, the techniques available to manage haemodynamic compromise, and the unique role which echocardiography plays in this complex process. CASE: A 69-year-old female presents to the emergency department with a fever, confusion and pain on urinating. Her blood pressure on arrival was 70/40, with heart rate of 117 bpm Despite 3 l of i.v. fluid she remained hypotensive. A central venous catheter was inserted and noradrenaline infusion commenced, and she was admitted to the intensive care unit for management of her shock state. At 6 h post admission, she was on high dose of noradrenaline (0.7 µg/kg per min) but blood pressure remained problematic. An echocardiogram was requested to better determine her haemodynamic state.