Using sequences of coronary angiograms to quantify the severity of stenosis

10/30/2024

Cardiovascular diseases (CVDs) are the leading cause of death globally, with Coronary Artery Disease (CAD) being the most prevalent contributor to CVD-related fatalities. Early diagnosis and precise assessment of stenosis severity are essential for improving patient outcomes and reducing the reliance on invasive, risky, and costly diagnostic procedures. This work develops a deep learning-based system that leverages angiogram videos to estimate Instantaneous Wave-Free Ratio (iFR) values, which are the gold standard metric for determining the need for surgical intervention in CAD patients. By utilizing advanced deep learning and computer vision techniques, this system seeks to provide accurate and automated quantification of stenosis, reducing human error, time, and cost associated with traditional methods. This approach has led to the development of a system that successfully estimates stenosis severity using angiogram sequences, demonstrating superior performance compared to benchmark models across most metrics. By integrating a ResNet-18 encoder, LSTM regressor, data augmentation, and auxiliary patient metadata, the system offers a novel, non-invasive approach to estimating the Instantaneous Wave-Free Ratio (iFR), aligning with current clinical standards. With further refinement and expansion of the annotated dataset, the system holds promise for clinical application, offering a more efficient, cost-effective, and accurate tool for Coronary Artery Disease (CAD) diagnosis and intervention planning.

Authors in the community:

• 

  M

  Mariana Serrão Guilherme

  ist1105045

Supervisors of this institution:

• 

  A

  Arlindo Manuel Limede de Oliveira

  ist12282

electrical-engineering-electronic-engineering-information-engineering - Electrical engineering, electronic engineering, information engineering

eng - English

Instituto Superior Técnico