PhD Scientific Days 2024

Budapest, 9-10 July 2024

Cardiovascular Medicine and Research I.

Automatic Coronary Artery Segmentation on Computed Tomography Coronary Angiography Images

Author(s)

Belén Serrano-Antón1,5,6, Alberto Otero-Cacho2,5,6, Diego López-Otero3,7, Brais Díaz-Fernández3,7, María Bastos Fernández3, Vicente Pérez-Muñuzuri4,8, José Ramón González-Juanatey3,7, Alberto P. Muñuzuri5,6, María Bastos-Fernández7
1: FlowReserve Labs S.L., Santiago de Compostela, Spain
2: FlowReserve Labs S.L., Santiago de Compostela, Spain.
3: Cardiology and Intensive Cardiac Care Department, University Hospital of Santiago de Compostela, Santiago de Compostela, Spain.
4: Group of Nonlinear Physics. University of Santiago de Compostela, Santiago de Compostela, Spain.
5: Galician Center for Mathematical Research and Technology (CITMAga), Santiago de Compostela, Spain.
6: Group of Nonlinear Physics. University of Santiago de Compostela, Santiago de Compostela, Spain
7: Centro de Investigación Biomédica en Red de Enfermedades Cardiovasculares (CIBERCV), Madrid, Spain
8: Institute CRETUS, Group of Nonlinear Physics, University of Santiago de Compostela, Santiago de Compostela, Spain

Text of the abstract

Introduction:
Coronary artery diseases pose a significant global health challenge, necessitating accurate assessment for effective management. However, traditional diagnostic methods relying on invasive procedures entail risks and limitations, emphasizing the need for non-invasive alternatives.

Aims:
Our research aims to address the shortcomings of invasive diagnostics by developing a non-invasive method to obtain 3D coronary artery geometries from computed tomography coronary angiography (CTCA). We further aim to apply artificial intelligence (AI) techniques to extract insights from CTCA data, enhancing diagnostic accuracy and reducing reliance on invasive interventions.

Method:
We utilize CTCA to capture detailed anatomical information of coronary vasculature, employing neural networks based on the UNet architecture and transfer learning strategies for robust segmentation of coronary arteries in three-dimensional space. This enables the simulation of fluid dynamics within coronary arteries to extract clinical parameters like fractional flow reserve (FFR).

Results:
Our approach yields accurate coronary artery geometries and facilitates parameter calculations without invasive techniques. Additionally, clinicians benefit from clear visualizations of the coronary tree, enhancing informed decision-making processes and ultimately improving patient care and outcomes.

Conclusion:
By overcoming the limitations of invasive diagnostics, our computational framework represents a significant advancement in cardiovascular diagnostics. We envision its continued refinement and validation to establish it as a cornerstone of next-generation precision medicine and patient-centered care.

Funding:
Financial support by the Spanish Ministerio de Economía y Competitividad and European Regional Development Fund, MCIN/AEI/ 10.13039/501100011033 and by “ERDF A way of making Europe”. Xunta de Galicia funded research under Research Grant No. 2021-PG036 and the Spanish Ministerio de Ciencia e Innovación MCIN/AEI/10.13039/501100011033 through the Industrial Doctorates Grant.