CathEF: Automated Assessment of Cardiac Systolic Function from Coronary Angiograms using AI



Robert Avram, Joshua P Barrios, Sean Abreau, Cheng Yee Goh, Zeeshan Ahmed, Kevin Chung, Derek Y So, Jeffrey E Olgin, Geoffrey H Tison


Coronary angiography is the standard clinical procedure that is performed for coronary heart disease, or whenever there is concern for a “heart attack,” and is central to nearly all related clinical decision making. Coronary heart disease is the leading cause of adult death worldwide.










Assessment of cardiac pumping function during the angiogram procedure is typically obtained through left ventriculography, an additional procedure requiring catheter insertion into the left ventricle and contrast injection, which confers additional risk.


We developed CathEF, a video-based deep neural network algorithm, to estimate left ventricular ejection fraction (LVEF) from standard coronary angiograms. LVEF is the standard clinical assessment of cardiac pump function and is valuable for patient management and treatment decisions.






Our work demonstrates that video-based AI can achieve fully-automated and accurate estimation of LVEF from standard routinely-obtained coronary angiograms of the left coronary artery. This provides an opportunity to estimate LVEF during nearly every angiogram in a noninvasive, risk-free manner.


Coronary angiography, sometimes called a left heart catheterization, is the critical minimally-invasive procedure that is performed in anyone for whom there is concern of coronary atherosclerotic disease, the same disease that blocks arteries and leads to “heart attacks.”

Understanding cardiac systolic (pump) function at the time of coronary angiography can be helpful to optimize clinical decision-making and treatment decisions, particularly in urgent settings such as during potentially life-threatening acute coronary syndromes (ACS). LVEF can be measured prior to angiography with an echocardiogram (ultrasound), but this is not always available and can be months or years old. 

During the angiogram procedure, LVEF can also be estimated by performing an additional procedure known as a left ventriculogram, where an additional catheter is inserted into the left ventricle and contrast dye is injected. However, this is not always performed in part because it carries increased risk and contrast exposure.



We sought to develop a novel approach to estimate cardiac systolic function without requiring additional procedures or increasing procedural risk.



To develop CathEF, we first analyzed real-world coronary angiograms using CathAI—a pipeline of multiple deep neural network algorithms—to automatically identify angiogram videos containing the left coronary artery (LCA) as their primary anatomic structure. 




Angiograms were paired with a corresponding transthoracic echocardiogram (TTE) assessment of LVEF performed either 3 months before or 1 month after the angiogram.



Our final UCSF dataset consisted of 26,087 LCA videos derived from 3,960 coronary angiograms and 3,404 distinct patients. This was then split into training, development and testing datasets to develop and internally validate CathAI. 


CathEF demonstrated strong performance to estimate low LVEF ≤40% in the UCSF test dataset. We then additionally validated it in an external validation dataset in real-world angiograms from the University of Ottawa Heart Institute, showing similarly strong performance.




In the UCSF test dataset (n=813), CathEF showed strong discrimination for LVEF ≤40% with an area under the receiver operating characteristic curve (AUC) of 0.911 (95% CI 0.887-0.934). We also tested discrimination for the cutoff of LVEF ≤50% and AUC=0.879 (95% CI 0.852-0.907). There were 22.7 greater odds of reduced LVEF in those that CathEF predicted LVEF ≤40%. Sensivitiy and specificity were 83.9% and 81.3%, respectively.


CathEF also estimated continuous LVEF percentages, comparing well against TTE LVEF. The median absolute error of CathEF versus TTE-LVEF was 8.5% (95% CI 8.1%-9.0%), and it was 3.7% (95% CI 2.7%-9.3%) when averaging CathEF's predictions across a patient's angiographic study. Intraclass correlation coefficient was 0.77.


CathEF remained consistent across sex, BMI, presence of obstructive coronary artery disease, left ventricular hypertrophy and reduced kidney function. This includes many patients for whom the additional contrast from left ventriculography can be harmful. CathEF performed similarly for patients presenting urgently for ACS as for non-ACS.


To examine external generalizability, we validated CathAI in real-world angiograms from the University of Ottawa Heart Institute, a separate healthcare institution in a different country. 


In 4,471 LCA videos derived from 776 angiograms from 744 patients paired with TTE-LVEF, CathEF achieved an AUC of 0.906 (95% CI: 0.881-0.931) to identify LVEF ≤40%. There were 27.3 greater odds of reduced LVEF in those that CathEF predicted LVEF ≤40% in the UOHI dataset. Sensitivity and specificity for LVEF ≤40% were 77.9% and 88.6%, respectively. 

When predicting continuous LVEF percentages, median absolute error was 7.0% compared to TTE-LVEF.



We used the guided GradCAM algorithm explainability techqniues to better understand how CathEF predicts LVEF from angiograms.

GradCAM consistently highlighted epicardial coronary arteries during cardiac systole, and largely not during diastole. This suggests that CathEF likely identifies patterns in coronary artery blood flow or movement during systole to predict LVEF.



Our work demonstrates that CathEF, a video-based neural network, can accurately estimate LVEF, a key measurement of cardiac pumping function, providing valuable information from standard coronary angiograms that is usually not accessible to clinicians, with no additional procedures or risk.

♦ CathEF exhibited high performance in both internal and external validation in real-world angiograms obtained from a health system in a different country.

♦ Performance remained robust in patients with varied demographics and medical comorbidities, particularly in kidney disease and ACS, populations that may benefit most from limiting contrast exposure.

♦ This represents a technological advancement to coronary angiography, offering a novel approach to estimate real-time LV systolic function from nearly any routine angiogram.



CathEF offers the opportunity to obtain real-time LVEF estimates from routine angiograms with no added risk or procedures. 

CathEF provides information from coronary angiography that is not usually accessible to clinicians and typically requires additional testing.

This technology has the potential to change the standard of care by providing real-time, dynamic assessment of cardiac systolic function during coronary angiography.



Email questions or inquiries here.