Video-based AI for beat-to-beat assessment of cardiac function.

Accurate assessment of cardiac function is crucial for the diagnosis of cardiovascular disease1, screening for cardiotoxicity2 and decisions regarding the clinical management of patients with a critical illness3. However, human assessment of cardiac function focuses on a limited sampling of cardiac cycles and has considerable inter-observer variability despite years of training4,5. Here, to overcome this challenge, we present a video-based deep learning algorithm-EchoNet-Dynamic-that surpasses the performance of human experts in the critical tasks of segmenting the left ventricle, estimating ejection fraction and assessing cardiomyopathy. Trained on echocardiogram videos, our model accurately segments the left ventricle with a Dice similarity coefficient of 0.92, predicts ejection fraction with a mean absolute error of 4.1% and reliably classifies heart failure with reduced ejection fraction (area under the curve of 0.97). In an external dataset from another healthcare system, EchoNet-Dynamic predicts the ejection fraction with a mean absolute error of 6.0% and classifies heart failure with reduced ejection fraction with an area under the curve of 0.96. Prospective evaluation with repeated human measurements confirms that the model has variance that is comparable to or less than that of human experts. By leveraging information across multiple cardiac cycles, our model can rapidly identify subtle changes in ejection fraction, is more reproducible than human evaluation and lays the foundation for precise diagnosis of cardiovascular disease in real time. As a resource to promote further innovation, we also make publicly available a large dataset of 10,030 annotated echocardiogram videos.

Pathogenesis of Giant Cell Arteritis and Takayasu Arteritis-Similarities and Differences.

PURPOSE OF REVIEW: Giant cell arteritis (GCA) and Takayasu arteritis (TAK) are auto-inflammatory and autoimmune diseases with a highly selective tissue tropism for medium and large arteries. In both diseases, CD4+ T cells and macrophages form granulomatous lesions within the arterial wall, a tissue site normally protected by immune privilege. Vascular lesions can be accompanied by an extravascular component, typically an intense hepatic acute phase response that produces well-known laboratory abnormalities, e.g., elevated ESR and CRP. It is unclear whether GCA and TAK lie on a spectrum of disease or whether they represent fundamentally different disease processes. RECENT FINDINGS: GCA and TAK share many clinical features, but there are substantial differences in genetics, epidemiology, disease mechanisms, response to treatment, and treatment complications that give rise to different disease trajectories. A significant difference lies in the composition of the wall-infiltrating immune cell compartment, which in TAK includes a significant population of CD8+ T cells as well as natural killer cells, specifying disparate disease effector pathways mediating tissue damage and vessel wall remodeling. Despite the similarities in tissue tropism and histomorphology, GCA and TAK are two distinct vasculitides that rely on separate disease mechanisms and require disease-specific approaches in diagnosis and management.

Focused transthoracic echocardiography during critical care medicine training: curriculum implementation and evaluation of proficiency*.

OBJECTIVES: We designed and implemented a focused transthoracic echocardiography curriculum for critical care medicine fellows participating in 1- and 2-year training programs. We quantitatively evaluated their proficiency in focused transthoracic echocardiography. DESIGN: Prospective study evaluating curriculum implementation and objective assessment of focused transthoracic echocardiography proficiency. SETTING: Medical and surgical ICUs at an academic teaching hospital. Simulation laboratory. SUBJECTS: Eighteen critical care medicine fellows. INTERVENTIONS: Training in focused transthoracic echocardiography followed by proficiency testing. MEASUREMENTS AND MAIN RESULTS: We assessed the ability of critical care medicine fellows to obtain and interpret focused transthoracic echocardiography images from critically ill patients and a from transthoracic echocardiography simulator. Using a cognitive examination test, we also evaluated each fellow's knowledge with regard to focused transthoracic echocardiography and each fellow's ability to interpret prerecorded focused transthoracic echocardiography images. After training, critical care medicine fellows were able to rapidly obtain five essential focused transthoracic echocardiography views: parasternal long axis, parasternal short axis, apical four chamber, subcostal four chamber, and subcostal inferior vena cava. Fellows were also able to expeditiously identify four important abnormalities: asystole, left ventricular dysfunction, right ventricular dilation and dysfunction, and a large pericardial effusion. CONCLUSIONS: A focused transthoracic echocardiography curriculum that includes quantitative measures of proficiency can be integrated into critical care medicine fellowship training programs.

High-Throughput Precision Phenotyping of Left Ventricular Hypertrophy With Cardiovascular Deep Learning.

IMPORTANCE: Early detection and characterization of increased left ventricular (LV) wall thickness can markedly impact patient care but is limited by under-recognition of hypertrophy, measurement error and variability, and difficulty differentiating causes of increased wall thickness, such as hypertrophy, cardiomyopathy, and cardiac amyloidosis. OBJECTIVE: To assess the accuracy of a deep learning workflow in quantifying ventricular hypertrophy and predicting the cause of increased LV wall thickness. DESIGN, SETTINGS, AND PARTICIPANTS: This cohort study included physician-curated cohorts from the Stanford Amyloid Center and Cedars-Sinai Medical Center (CSMC) Advanced Heart Disease Clinic for cardiac amyloidosis and the Stanford Center for Inherited Cardiovascular Disease and the CSMC Hypertrophic Cardiomyopathy Clinic for hypertrophic cardiomyopathy from January 1, 2008, to December 31, 2020. The deep learning algorithm was trained and tested on retrospectively obtained independent echocardiogram videos from Stanford Healthcare, CSMC, and the Unity Imaging Collaborative. MAIN OUTCOMES AND MEASURES: The main outcome was the accuracy of the deep learning algorithm in measuring left ventricular dimensions and identifying patients with increased LV wall thickness diagnosed with hypertrophic cardiomyopathy and cardiac amyloidosis. RESULTS: The study included 23 745 patients: 12 001 from Stanford Health Care (6509 [54.2%] female; mean [SD] age, 61.6 [17.4] years) and 1309 from CSMC (808 [61.7%] female; mean [SD] age, 62.8 [17.2] years) with parasternal long-axis videos and 8084 from Stanford Health Care (4201 [54.0%] female; mean [SD] age, 69.1 [16.8] years) and 2351 from CSMS (6509 [54.2%] female; mean [SD] age, 69.6 [14.7] years) with apical 4-chamber videos. The deep learning algorithm accurately measured intraventricular wall thickness (mean absolute error [MAE], 1.2 mm; 95% CI, 1.1-1.3 mm), LV diameter (MAE, 2.4 mm; 95% CI, 2.2-2.6 mm), and posterior wall thickness (MAE, 1.4 mm; 95% CI, 1.2-1.5 mm) and classified cardiac amyloidosis (area under the curve [AUC], 0.83) and hypertrophic cardiomyopathy (AUC, 0.98) separately from other causes of LV hypertrophy. In external data sets from independent domestic and international health care systems, the deep learning algorithm accurately quantified ventricular parameters (domestic: R2, 0.96; international: R2, 0.90). For the domestic data set, the MAE was 1.7 mm (95% CI, 1.6-1.8 mm) for intraventricular septum thickness, 3.8 mm (95% CI, 3.5-4.0 mm) for LV internal dimension, and 1.8 mm (95% CI, 1.7-2.0 mm) for LV posterior wall thickness. For the international data set, the MAE was 1.7 mm (95% CI, 1.5-2.0 mm) for intraventricular septum thickness, 2.9 mm (95% CI, 2.4-3.3 mm) for LV internal dimension, and 2.3 mm (95% CI, 1.9-2.7 mm) for LV posterior wall thickness. The deep learning algorithm accurately detected cardiac amyloidosis (AUC, 0.79) and hypertrophic cardiomyopathy (AUC, 0.89) in the domestic external validation site. CONCLUSIONS AND RELEVANCE: In this cohort study, the deep learning model accurately identified subtle changes in LV wall geometric measurements and the causes of hypertrophy. Unlike with human experts, the deep learning workflow is fully automated, allowing for reproducible, precise measurements, and may provide a foundation for precision diagnosis of cardiac hypertrophy.

Association of Thoracic Aortic Aneurysm Size With Long-term Patient Outcomes: The KP-TAA Study.

Importance: The risk of adverse events from ascending thoracic aorta aneurysm (TAA) is poorly understood but drives clinical decision-making. Objective: To evaluate the association of TAA size with outcomes in nonsyndromic patients in a large non-referral-based health care delivery system. Design, Setting, and Participants: The Kaiser Permanente Thoracic Aortic Aneurysm (KP-TAA) cohort study was a retrospective cohort study at Kaiser Permanente Northern California, a fully integrated health care delivery system insuring and providing care for more than 4.5 million persons. Nonsyndromic patients from a regional TAA safety net tracking system were included. Imaging data including maximum TAA size were merged with electronic health record (EHR) and comprehensive death data to obtain demographic characteristics, comorbidities, medications, laboratory values, vital signs, and subsequent outcomes. Unadjusted rates were calculated and the association of TAA size with outcomes was evaluated in multivariable competing risk models that categorized TAA size as a baseline and time-updated variable and accounted for potential confounders. Data were analyzed from January 2018 to August 2021. Exposures: TAA size. Main Outcomes and Measures: Aortic dissection (AD), all-cause death, and elective aortic surgery. Results: Of 6372 patients with TAA identified between 2000 and 2016 (mean [SD] age, 68.6 [13.0] years; 2050 female individuals [32.2%] and 4322 male individuals [67.8%]), mean (SD) initial TAA size was 4.4 (0.5) cm (828 individuals [13.0% of cohort] had initial TAA size 5.0 cm or larger and 280 [4.4%] 5.5 cm or larger). Rates of AD were low across a mean (SD) 3.7 (2.5) years of follow-up (44 individuals [0.7% of cohort]; incidence 0.22 events per 100 person-years). Larger initial aortic size was associated with higher risk of AD and all-cause death in multivariable models, with an inflection point in risk at 6.0 cm. Estimated adjusted risks of AD within 5 years were 0.3% (95% CI, 0.3-0.7), 0.6% (95% CI, 0.4-1.3), 1.5% (95% CI, 1.2-3.9), 3.6% (95% CI, 1.8-12.8), and 10.5% (95% CI, 2.7-44.3) in patients with TAA size of 4.0 to 4.4 cm, 4.5 to 4.9 cm, 5.0 to 5.4 cm, 5.5 to 5.9 cm, and 6.0 cm or larger, respectively, in time-updated models. Rates of the composite outcome of AD and all-cause death were higher than for AD alone, but a similar inflection point for increased risk was observed at 6.0 cm. Conclusions and Relevance: In a large sociodemographically diverse cohort of patients with TAA, absolute risk of aortic dissection was low but increased with larger aortic sizes after adjustment for potential confounders and competing risks. Our data support current consensus guidelines recommending prophylactic surgery in nonsyndromic individuals with TAA at a 5.5-cm threshold.

Coronary Artery Vasospasm Requiring Cardiac Autotransplantation Yet Controlled With Tobacco.

Coronary artery vasospasm is typically managed through avoidance of triggers and with symptomatic treatments with calcium channel blockers and long-acting nitrates. Here, we report a rare case of medically refractory coronary artery vasospasm associated with genetic predispositions that initially required cardiac autotransplantation followed paradoxically by nicotine for long-term symptomatic control. (Level of Difficulty: Intermediate.).

Deep learning evaluation of biomarkers from echocardiogram videos.

BACKGROUND: Laboratory testing is routinely used to assay blood biomarkers to provide information on physiologic state beyond what clinicians can evaluate from interpreting medical imaging. We hypothesized that deep learning interpretation of echocardiogram videos can provide additional value in understanding disease states and can evaluate common biomarkers results. METHODS: We developed EchoNet-Labs, a video-based deep learning algorithm to detect evidence of anemia, elevated B-type natriuretic peptide (BNP), troponin I, and blood urea nitrogen (BUN), as well as values of ten additional lab tests directly from echocardiograms. We included patients (n = 39,460) aged 18 years or older with one or more apical-4-chamber echocardiogram videos (n = 70,066) from Stanford Healthcare for training and internal testing of EchoNet-Lab's performance in estimating the most proximal biomarker result. Without fine-tuning, the performance of EchoNet-Labs was further evaluated on an additional external test dataset (n = 1,301) from Cedars-Sinai Medical Center. We calculated the area under the curve (AUC) of the receiver operating characteristic curve for the internal and external test datasets. FINDINGS: On the held-out test set of Stanford patients not previously seen during model training, EchoNet-Labs achieved an AUC of 0.80 (0.79-0.81) in detecting anemia (low hemoglobin), 0.86 (0.85-0.88) in detecting elevated BNP, 0.75 (0.73-0.78) in detecting elevated troponin I, and 0.74 (0.72-0.76) in detecting elevated BUN. On the external test dataset from Cedars-Sinai, EchoNet-Labs achieved an AUC of 0.80 (0.77-0.82) in detecting anemia, of 0.82 (0.79-0.84) in detecting elevated BNP, of 0.75 (0.72-0.78) in detecting elevated troponin I, and of 0.69 (0.66-0.71) in detecting elevated BUN. We further demonstrate the utility of the model in detecting abnormalities in 10 additional lab tests. We investigate the features necessary for EchoNet-Labs to make successful detection and identify potential mechanisms for each biomarker using well-known and novel explainability techniques. INTERPRETATION: These results show that deep learning applied to diagnostic imaging can provide additional clinical value and identify phenotypic information beyond current imaging interpretation methods. FUNDING: J.W.H. and B.H. are supported by the NSF Graduate Research Fellowship. D.O. is supported by NIH K99 HL157421-01. J.Y.Z. is supported by NSF CAREER 1942926, NIH R21 MD012867-01, NIH P30AG059307 and by a Chan-Zuckerberg Biohub Fellowship.

Cellular Signaling Pathways in Medium and Large Vessel Vasculitis.

Autoimmune and autoinflammatory diseases of the medium and large arteries, including the aorta, cause life-threatening complications due to vessel wall destruction but also by wall remodeling, such as the formation of wall-penetrating microvessels and lumen-stenosing neointima. The two most frequent large vessel vasculitides, giant cell arteritis (GCA) and Takayasu arteritis (TAK), are HLA-associated diseases, strongly suggestive for a critical role of T cells and antigen recognition in disease pathogenesis. Recent studies have revealed a growing spectrum of effector functions through which T cells participate in the immunopathology of GCA and TAK; causing the disease-specific patterning of pathology and clinical outcome. Core pathogenic features of disease-relevant T cells rely on the interaction with endothelial cells, dendritic cells and macrophages and lead to vessel wall invasion, formation of tissue-damaging granulomatous infiltrates and induction of the name-giving multinucleated giant cells. Besides antigen, pathogenic T cells encounter danger signals in their immediate microenvironment that they translate into disease-relevant effector functions. Decisive signaling pathways, such as the AKT pathway, the NOTCH pathway, and the JAK/STAT pathway modify antigen-induced T cell activation and emerge as promising therapeutic targets to halt disease progression and, eventually, reset the immune system to reestablish the immune privilege of the arterial wall.

Innate and Adaptive Immunity in Giant Cell Arteritis.

Autoimmune diseases can afflict every organ system, including blood vessels that are critically important for host survival. The most frequent autoimmune vasculitis is giant cell arteritis (GCA), which causes aggressive wall inflammation in medium and large arteries and results in vaso-occlusive wall remodeling. GCA shares with other autoimmune diseases that it occurs in genetically predisposed individuals, that females are at higher risk, and that environmental triggers are suspected to beget the loss of immunological tolerance. GCA has features that distinguish it from other autoimmune diseases and predict the need for tailored diagnostic and therapeutic approaches. At the core of GCA pathology are CD4+ T cells that gain access to the protected tissue niche of the vessel wall, differentiate into cytokine producers, attain tissue residency, and enforce macrophages differentiation into tissue-destructive effector cells. Several signaling pathways have been implicated in initiating and sustaining pathogenic CD4+ T cell function, including the NOTCH1-Jagged1 pathway, the CD28 co-stimulatory pathway, the PD-1/PD-L1 co-inhibitory pathway, and the JAK/STAT signaling pathway. Inadequacy of mechanisms that normally dampen immune responses, such as defective expression of the PD-L1 ligand and malfunction of immunosuppressive CD8+ T regulatory cells are a common theme in GCA immunopathology. Recent studies are providing a string of novel mechanisms that will permit more precise pathogenic modeling and therapeutic targeting in GCA and will fundamentally inform how abnormal immune responses in blood vessels lead to disease.

Deep learning interpretation of echocardiograms.

Echocardiography uses ultrasound technology to capture high temporal and spatial resolution images of the heart and surrounding structures, and is the most common imaging modality in cardiovascular medicine. Using convolutional neural networks on a large new dataset, we show that deep learning applied to echocardiography can identify local cardiac structures, estimate cardiac function, and predict systemic phenotypes that modify cardiovascular risk but not readily identifiable to human interpretation. Our deep learning model, EchoNet, accurately identified the presence of pacemaker leads (AUC = 0.89), enlarged left atrium (AUC = 0.86), left ventricular hypertrophy (AUC = 0.75), left ventricular end systolic and diastolic volumes ( R 2 = 0.74 and R 2 = 0.70), and ejection fraction ( R 2 = 0.50), as well as predicted systemic phenotypes of age ( R 2 = 0.46), sex (AUC = 0.88), weight ( R 2 = 0.56), and height ( R 2 = 0.33). Interpretation analysis validates that EchoNet shows appropriate attention to key cardiac structures when performing human-explainable tasks and highlights hypothesis-generating regions of interest when predicting systemic phenotypes difficult for human interpretation. Machine learning on echocardiography images can streamline repetitive tasks in the clinical workflow, provide preliminary interpretation in areas with insufficient qualified cardiologists, and predict phenotypes challenging for human evaluation.

Do annuloplasty rings designed to treat ischemic/functional mitral regurgitation alter left-ventricular dimensions in the acutely ischemic ovine heart?

OBJECTIVE: To quantify the effects of annuloplasty rings designed to treat ischemic/functional mitral regurgitation on left ventricular septal-lateral (S-L) and commissure-commissure (C-C) dimensions. METHODS: Radiopaque markers were placed as opposing pairs on the S-L and C-C aspects of the mitral annulus and the basal, equatorial, and apical level of the left ventricle (LV) in 30 sheep. Ten true-sized Carpentier-Edwards Physio (PHY), Edwards IMR ETlogix (ETL), and GeoForm (GEO; all from Edwards Lifesciences, Irvine, Calif) annuloplasty rings were inserted in a releasable fashion. After 90 seconds of left circumflex artery occlusion with the ring implanted (RING), 4-dimensional marker coordinates were obtained using biplane videofluoroscopy. After ring release, another data set was acquired after another 90 seconds of left circumflex artery occlusion (NO RING). S-L and C-C diameters were computed as the distances between the respective marker pairs at end-diastole. Percent change in diameters was calculated between RING versus NO RING as 100 × (diameter in centimeters [RING] - diameter in centimeters [NO RING])/diameter in centimeters [NO RING]). RESULTS: Compared with NO RING, all ring types (PHY, ETL, and GEO) reduced mitral annular S-L dimensions by -20.7 ± 5.6%, -26.8 ± 3.9%, and -34.5 ± 3.8%, respectively. GEO reduced the S-L dimensions of the LV at the basal level only by -2.3 ± 2.4%, whereas all other S-L dimensions of the LV remained unchanged with all 3 rings implanted. PHY, ETL, and GEO reduced mitral annular C-C dimensions by -17.5 ± 4.8%, -19.6 ± 2.5, and -8.3 ± 4.9%, respectively, but none of the rings altered the C-C dimensions of the LV. CONCLUSIONS: Despite radical reduction of mitral annular size, disease-specific ischemic/functional mitral regurgitation annuloplasty rings do not induce relevant changes of left ventricular dimensions in the acutely ischemic ovine heart.

Advances in echocardiography.

Echocardiography remains the main tool for noninvasive cardiac evaluation. Advances in echo technology and techniques offer new information, which will impact both the timing and method of surgical intervention. Three-dimensional echocardiography, in particular, provides improved tools for quantification both of volumes and of flows. Geometrical relations necessary for understanding functional abnormalities are also preserved with three-dimensional (3D) echocardiography. Finally 3D echocardiography also provides a unique tool for guiding minimally invasive interventions.

Technical advances in cardiovascular imaging.

Cardiovascular imaging technology is continuously evolving and provides an increasing array of tests to evaluate cardiovascular morphology and function. A basic understanding of imaging technology is helpful to select the best modality to answer a specific clinical question. This article provides a brief overview of recent technical developments in computed tomography (CT), magnetic resonance (MR), and echocardiography, which have increased our diagnostic understanding and may modulate treatment planning of patients with cardiovascular diseases: electrocardiographically (ECG)-gated CT, 4D-flow magnetic resonance imaging (MRI), and three-dimensional (3D) echocardiography.

Pre- and postoperative imaging of the aortic root for valve-sparing aortic root repair (V-SARR).

Valve-sparing aortic root repair (V-SARR) using the David reimplantation method is an increasingly popular alternative to composite valve graft aortic root replacement in patients with aortic root aneurysms or dissections who wish to avoid anticoagulation. Computed tomography (CT) with retrospective electrocardiograph (ECG)-gating has become routine before and following V-SARR at Stanford. CT allows accurate measurement of aortic dimensions and provides unprecedented three-dimensional (3D) images of the sinuses, the aortic valve cusps, and coronary arteries in patients with the Marfan syndrome (MFS), with a bicuspid aortic valve (BAV), or other aortic diseases. This helps the surgeon to conceptualize the size of the aortic grafts required and how much reduction is necessary proximally (aortic annulus) and distally. These maneuvers are used to reduce the aortic annular diameter (when necessary) and replace the sinuses and ascending aorta (T. David-V, Stanford modification V-SARR). Postoperative ECG-gated CT confirms the reconstructed geometry and reliably detects coronary or other anastomotic problems.

Laser-Machined Shape Memory Alloy Sensors for Position Feedback in Active Catheters.

Catheter-based interventions are a form of minimally invasive surgery that can decrease hospitalization time and greatly lower patient morbidity compared to traditional methods. However, percutaneous catheter procedures are hindered by a lack of precise tip manipulation when actuation forces are transmitted over the length of the catheter. Active catheters with local shape-memory-alloy (SMA) actuation can potentially provide the desired manipulation of a catheter tip, but hysteresis makes it difficult to control the actuators. A method to integrate small-volume, compliant sensors on an active catheter to provide position feedback for control would greatly improve the viability of SMA-based active catheters. In this work, we describe the design, fabrication, and performance of resistance-based position sensors that are laser-machined from superelastic SMA tubing. Combining simple material models and rapid prototyping, we can develop sensors of appropriate stiffness and sensitivity with simple modifications in sensor geometry. The sensors exhibit excellent linearity over the operating range and are designed to be easily integrated onto an active catheter substrate.