Book chapter
Chapter 8 - Quantitative Assessment and Prediction of Coronary Plaque Development Using Serial Intravascular Ultrasound and Virtual Histology
Intravascular Ultrasound, pp.121-140
Elsevier Ltd
2020
DOI: 10.1016/B978-0-12-818833-0.00008-4
Abstract
The mechanisms of plaque development are still not entirely understood. Intravascular ultrasound (IVUS) is a preferred modality to assess complex plaque structures in coronary arteries. Virtual histology (VH) determines one of four plaque types for each coronary wall location. Serial imaging over multiple time points assesses plaque development in terms of extent and severity. Identification of arterial locations likely to later develop high-risk plaques is highly desirable to prevent future major adverse cardiac events by targeted monitoring and intervention. Ultimately, we aim to predict future plaque development from just the VH-IVUS and systemic/biomarker information available at the time of baseline imaging. Processing starts with the segmentation of lumen and adventitia borders from the IVUS pullbacks using our established multidimensional optimal-graph search LOGISMOS framework with efficient “Just Enough Interaction” editing, substantially reducing time and manual effort to obtain accurate borders. IVUS registration is a challenging problem due to inconsistent pullback start and end points, irregularities in IVUS catheter speed, longitudinal gating artifacts, artifactual angular twisting, and changes in vessel morphology over time. We have developed a novel method for simultaneous registration of location and orientation of two IVUS pullbacks of the same vessel from different time points, using 3D graph optimization. For the prediction of future high-risk coronary plaque locations and types, we have developed and validated a comprehensive machine-learning method based on well-accepted criteria (thin-cap fibroatheroma, plaque burden ≥70%, or minimal luminal area ≤4mm2) to define such risk. Our method utilizes 236 image-location and 18 patient-specific features for seven predictors of the individual vessel morphology and higher-level plaque-severity features, and implements a subsequent per-predictor feature-selection process to reduce complexity.
Details
- Title: Subtitle
- Chapter 8 - Quantitative Assessment and Prediction of Coronary Plaque Development Using Serial Intravascular Ultrasound and Virtual Histology
- Creators
- Andreas Wahle - University of IowaLing Zhang - University of IowaZhi Chen - University of IowaHonghai Zhang - University of IowaJohn J Lopez - Loyola University ChicagoTomas Kovarnik - Charles UniversityMilan Sonka - University of Iowa
- Resource Type
- Book chapter
- Publication Details
- Intravascular Ultrasound, pp.121-140
- DOI
- 10.1016/B978-0-12-818833-0.00008-4
- Publisher
- Elsevier Ltd
- Language
- English
- Date published
- 2020
- Academic Unit
- Roy J. Carver Department of Biomedical Engineering; Electrical and Computer Engineering; Radiation Oncology; The Iowa Institute for Biomedical Imaging; Fraternal Order of Eagles Diabetes Research Center; Injury Prevention Research Center; Ophthalmology and Visual Sciences
- Record Identifier
- 9984186913402771
Metrics
115 Record Views