Accurate classification of plaque composition is essential for treatment planning. Deep learning (DL) methods have been introduced for this purpose, to analyze intravascular images and characterize in a fast and subjective manner plaque types. In this study, we compared the efficacy of two DL methods, designed to process data acquired by two intravascular–an optical coherence tomography (OCT) and a near-infrared spectroscopy-intravascular ultrasound (NIRS-IVUS)–catheters to assess plaque types using histology as the reference standard. We matched histology, OCT, and NIRS-IVUS images, compared their estimations, and found that the DL method developed for NIRS-IVUS analysis had a better correlation with histology for calcific and lipidic tissue as compared with the OCT-DL method while both methods had a moderate correlation with the estimations of histology for fibrotic tissue. These findings could be attributed to the fact that OCT due to its poor penetration especially in lesions with large plaque burden fails to identify the deep-seated plaque and also to the fact that the NIRS-IVUS-DL method was developed with the use of histology instead of experts’ analysis.
Combined intravascular ultrasound-optical coherence tomography (IVUS-OCT) enables more accurate coronary plaque tissue classification compared to single modality systems. Automated solutions are needed to that take advantage of information from both modalities to speed such analysis. This study aimed to train and validate a deep learning (DL) model for tissue classification in combined IVUS-OCT images. Coronary segments from 8 arteries from cadaveric human hearts were studied with the Novasight Hybrid imaging catheter. IVUS-OCT images were matched with histological sections and tissue types annotated. These regions of interest were used train and test a DL-classifier for plaque composition (949 matched histological and IVUS-OCT frames from 8 patients for training, 306 frames from 2 patients for testing). The accuracy of the classifier for regional classification was 78.8% suggesting that the trained DL-model is capable of accurate tissue type classification in combined IVUS-OCT images.
Access to the requested content is limited to institutions that have purchased or subscribe to SPIE eBooks.
You are receiving this notice because your organization may not have SPIE eBooks access.*
*Shibboleth/Open Athens users─please
sign in
to access your institution's subscriptions.
To obtain this item, you may purchase the complete book in print or electronic format on
SPIE.org.
INSTITUTIONAL Select your institution to access the SPIE Digital Library.
PERSONAL Sign in with your SPIE account to access your personal subscriptions or to use specific features such as save to my library, sign up for alerts, save searches, etc.