Search

BackgroundAutomatic coronary angiography (CAG) assessment may help in faster screening and diagnosis of stenosis in patients with atherosclerotic disease. We aimed to provide an end-to-end workflow that separates cases with normal or mild stenoses from those with higher stenosis severities to facilitate safety screening of a large volume of the CAG images.MethodsA deep learning-based end-to-end workflow was employed as follows: (1) Candidate frame selection from CAG videograms with Convolutional Neural Network (CNN) + Long Short Term Memory (LSTM) network, (2) Stenosis classification with Inception-v3 using 2 or 3 categories (25%, and/or total occlusion) with and without redundancy training, and (3) Stenosis localization with two methods of class activation map (CAM) and anchor-based feature pyramid network (FPN). Overall 13,744 frames from 230 studies were used for the stenosis classification training and fourfold cross-validation for image-, artery-, and per-patient-level. For the stenosis localization training and fourfold cross-validation, 690 images with > 25% stenosis were used.ResultsOur model achieved an accuracy of 0.85, sensitivity of 0.96, and AUC of 0.86 in per-patient level stenosis classification. Redundancy training was effective to improve classification performance. Stenosis position localization was adequate with better quantitative results in anchor-based FPN model, achieving global-sensitivity for left coronary artery (LCA) and right coronary artery (RCA) of 0.68 and 0.70.ConclusionWe demonstrated a fully automatic end-to-end deep learning-based workflow that eliminates the vessel extraction and segmentation step in coronary artery stenosis classification and localization on CAG images. This tool may be useful to facilitate safety screening in high-volume centers and in clinical trial settings.

BackgroundThere are multiple predictive factors for cardiovascular (CV) mortality measured at, or after heart failure (HF) diagnosis. However, the predictive role of long-term exposure to these predictors prior to HF diagnosis is unknown.ObjectivesWe aim to identify predictive factors of CV mortality in participants with HF, using cumulative exposure to risk factors before HF development.MethodsParticipants of Multi-Ethnic Study of Atherosclerosis (MESA) with incident HF were included. We used stepwise Akaike Information Criterion to select CV mortality predictors among clinical, biochemical, and imaging markers collected prior to HF. Using the AUC of B-spline-corrected curves, we estimated cumulative exposure to predictive factors from baseline to the last exam before HF. The prognostic performance for CV mortality after HF was evaluated using competing risk regression with non-CV mortality as the competing risk.ResultsOverall, 375 participants had new HF events (42.9% female, mean age: 74). Over an average follow-up of 4.7 years, there was no difference in the hazard of CV death for HF with reduced versus preserved ejection fraction (HR = 1.27, p = 0.23). The selected predictors of CV mortality in models with the least prediction error were age, cardiac arrest, myocardial infarction, and diabetes, QRS duration, HDL, cumulative exposure to total cholesterol and glucose, NT-proBNP, left ventricular mass, and statin use. The AUC of the models were 0.72 when including the latest exposure to predictive factors and 0.79 when including cumulative prior exposure to predictive factors (p = 0.20).ConclusionIn HF patients, besides age and diagnosed diabetes or CVD, prior lipid profile, NT-proBNP, LV mass, and QRS duration available at the diagnosis time strongly predict CV mortality. Implementing cumulative exposure to cholesterol and glucose, instead of latest measures, improves predictive accuracy for HF mortality, though not reaching statistical significance.