A preliminary study of IVOCT-based atherosclerosis plaque classification technique
Document Type
Conference Item
Publication Date
1-1-2022
Abstract
Atherosclerosis is a type of cardiovascular disease (CVD) that affects the coronary artery by build-up of plaque, which can potentially cause stroke or ischemic damage to the surrounding tissue. Intravascular Optical Coherence Tomography (IVOCT), an imaging modality, is able to capture detailed images of arteries affected by atherosclerosis that contain identifiable characteristics. These characteristics can assist clinicians to differentiate certain plaque types such as, fibrous, calcific and lipid, and provide diagnosis appropriately. However, clinicians face challenges in manual visual plaque identification from IVOCT images such as fatigue and IVOCT artifacts. Hence, the aim of this study is to produce an automated IVOCT-based plaque segmentation method to assist clinicians in their diagnosis. This preliminary study investigated only two plaque types, which are fibrous and calcified plaque as they are much more prominent to be labelled manually. The image dataset was pre-processed with Gabor filters before training the Random Forest (RF) and XGBoost models. The results demonstrated that the XGBoost model performed slightly better than the Random Forest model with 82.0 and 80.9 accuracy respectively. This shows that machine learning techniques can be applied conveniently to assist, automate and reduce the time for clinician’s visual assessment in the overall diagnosis workflow. © 2022, Springer Nature Switzerland AG.
Keywords
Atherosclerosis, Image segmentation, Intravascular optical coherence tomography, Machine learning, Plaque classification
Divisions
biomedengine,medicinedept
Funders
Malaysia Ministry of Higher Education Fundamental Research Grant Scheme [Grant no. FRGS/1/2018/SKK03/UM/02/1, GPF026A-2019]
Publication Title
IFMBE Proceedings
Volume
86
Publisher
Springer Science and Business Media Deutschland GmbH
Event Title
6th Kuala Lumpur International Conference on Biomedical Engineering, BioMed 2021
Event Location
Virtual, Online
Event Dates
28-29 July 2021
Event Type
conference