AUTHOR=Huang Jiayue , Ninomiya Kai , Tu Shengxian , Masuda Shinichiro , Dijkstra Jouke , Chu Miao , Ding Daixin , Hynes Sean O. , O’Leary Neil , Wijns William , Onuma Yoshinobu , Serruys Patrick W.
TITLE=Calcified plaque detected on OCT with deep learning and cross-validated with optical and ultrasound signals: A complementary appraisal and preamble to combined IVUS-OCT catheter
JOURNAL=Frontiers in Photonics
VOLUME=3
YEAR=2022
URL=https://www.frontiersin.org/journals/photonics/articles/10.3389/fphot.2022.1019552
DOI=10.3389/fphot.2022.1019552
ISSN=2673-6853
ABSTRACT=
Background: The optical coherence tomography (OCT)-deep learning (DL) emerged as a promising tool for automated plaque characterization. However, its findings have not been compared with optical and ultrasound signals.
Objectives: The objective was to cross-validate the calcified plaque detected by OCT-DL, using comprehensive tissue characterization technologies including OCT-derived optical properties, intravascular ultrasound (IVUS)-virtual histology (VH) and echogenicity.
Methods: Five years after bioresorbable scaffold (BRS) implantation, 15 patients underwent OCT and IVUS imaging. The unique platinum markers in BRS facilitated precise OCT-IVUS co-registration. Calcified plaque detected by OCT-DL were corroborated by/with optical properties, greyscale (GS)-IVUS, IVUS-VH and echogenicity. The concordance between OCT-DL and other modalities was assessed by kappa statistics. The calcium arc measured by different modalities were compared by orthogonal linear regression.
Results: Forty-three calcified plaques were detected by DL in 72 matched anatomic slices, 41 (95%) were confirmed as pure (n = 29) or hybrid calcified plaque (n = 12) by optical properties. Weighted kappa between OCT-DL and GS-IVUS, IVUS-VH and echogenicity were 0.69, 0.60 and 0.60, respectively. After having excluded artifactual optical shadowing (n = 5) generated by guidewire or platinum marker, kappa increased to 0.77, 0.68 and 0.69, with agreement ranging between 90% and 93%. Calcium arc derived from OCT-DL showed moderate correlation and agreement with GS-IVUS (ICCa = 0.81, difference = 1.73 ± 15.25°), IVUS-VH (ICCa = 0.69, difference = -5.60 ± 21.19°) and echogenicity (ICCa = 0.65, difference = 10.28 ± 18.70°).
Conclusion: OCT empowered by deep learning showed substantial agreement with optical and ultrasound signals. The comprehensive assessment provided by OCT and IVUS heralds the potential diagnostic value of combined IVUS-OCT catheters.