Share Email Print

Proceedings Paper

Deep learning segmentation used in IVOCT images to guide optical attenuation imaging for plaque characterization (Conference Presentation)

Paper Abstract

Light attenuation has been used for a better understanding of plaque build-up in coronary arteries. The current analysis is only useful in diseased segments. We applied an automated detection using a deep-learning approach to identify the diseased areas. A U-net was trained to detect the lumen, the guide-wire structure, healthy vessel wall, and the diseased vessel wall. The trained network achieves an average Dice index of 0.88±0.02. Applying it to all images of the testing pullbacks, diseased areas were segmented. The attenuation was estimated in this area and can be visualized in a 3-D view reconstructed using the detected lumen regions.

Paper Details

Date Published: 9 March 2020
Proc. SPIE 11215, Diagnostic and Therapeutic Applications of Light in Cardiology 2020, 1121506 (9 March 2020); doi: 10.1117/12.2545639
Show Author Affiliations
Shengnan Liu, Erasmus MC (Netherlands)
Denis Shamonin, Leiden Univ. Medical Ctr. (Netherlands)
Guillaume Zahnd, Technische Univ. München (Germany)
Joost Daemen, Erasmus MC (Netherlands)
A. F. W. van der Steen, Erasmus MC (Netherlands)
Theo van Walsum, Erasmus MC (Netherlands)
Gijs van Soest, Erasmus MC (Netherlands)

Published in SPIE Proceedings Vol. 11215:
Diagnostic and Therapeutic Applications of Light in Cardiology 2020
Kenton W. Gregory M.D.; Laura Marcu, Editor(s)

© SPIE. Terms of Use
Back to Top
Sign in to read the full article
Create a free SPIE account to get access to
premium articles and original research
Forgot your username?