In:
Circulation, Ovid Technologies (Wolters Kluwer Health), Vol. 146, No. Suppl_1 ( 2022-11-08)
Abstract:
Introduction: Intravascular optical coherence tomography (OCT) provides a high-resolution representation of coronary atherosclerosis. However, the accuracy of OCT image interpretation is unknown. We present a blinded study of histologically identified OCT human plaque composition by leading international OCT core labs. Hypothesis: Human readers are unable to accurately identify all plaque compositions. Methods: Seven recognized OCT core labs (Japan, Netherlands, Switzerland, USA) participated. Each laboratory was provided 51 OCT image clips recorded from ex vivo human coronary arteries within 24 hours of death and had regions of interest processed for H & E stain. Nine plaque components were identified and are listed in Results. Each OCT core lab was asked to blindly read OCT images and complete a questionnaire indicating one of nine possible plaque types. Kappa coefficient ( k ) was used ( k 〈 0.5 poor agreement, k 〉 0.5 high agreement) to measure intra-reader agreement; the median k is reported for each plaque component. Results: High agreement was demonstrated in identifying stable plaque including fibrous (0.93), calcium (0.83), layered plaque (1.0), and ThCFA (0.63), in contrast to unstable plaque including calcified nodule (0.50), lipid pool (0.35), macrophages/foam cells (0.35), TCFA (0.39), and necrotic core (0.22). When categories were restricted to Fibrous, Calcium, and Lipids, with Lipids expanded to include lipid pool, TCFA, and necrotic core, similar to an OCT core lab definition, k for Lipids increased to 0.61. Conclusions: Results suggest that participating OCT core labs can accurately identify combined lipid categories (lipid pool, TCFA, and necrotic core), fibrous tissue and calcium. However, accuracy is reduced when reading more sophisticated unstable plaque components. Study results emphasize the need for the development of OCT plaque composition analysis with artificial intelligence.
Type of Medium:
Online Resource
ISSN:
0009-7322
,
1524-4539
DOI:
10.1161/circ.146.suppl_1.11941
Language:
English
Publisher:
Ovid Technologies (Wolters Kluwer Health)
Publication Date:
2022
detail.hit.zdb_id:
1466401-X
