Stroke is a fatal disease in developing countries. Every year millions of people die due to stroke and other cardiovascular diseases (CVD). Main cause of CVD and stroke is atherosclerosisAtherosclerosis disease. Modern medical treatments include automated detection of atherosclerotic plaque tissue detection in ultrasound images. Imaging plaques using ultrasound avoids radiation is inexpensive and readily available in most diagnostic centers. Image analysis can assist in stratifying stroke risk using plaque tissue characterization (PTC). Earlier strategies have used machine learningMachine Learning (ML)-based approaches but are ad-hoc, tedious, and not accurate. We presented two deep learning modelsDeep Learning Models for internal carotid artery (ICA) plaque segmentation. The methodology consisted of design of two deep learning models,Deep Learning Models named UNetUNetand SegNet-UNetSegNet-UNet. Each of the models had a depth of four layers that controlled the number of parameters and computation time. We considered a database of 970 B-mode ICA ultrasound images of high-risk patients. Also, we used a K10 cross-validation on complete database. Finally, we calculated the plaque area (in mm2) using the segmented pixels. We achieved correlation coefficient of 0.98 (p-value < 0.001) between estimated and ground truth plaque area for both models. The overall system performs segmentation of one image in fraction of a second.
Hybrid Deep Learning Models for Segmentation of Atherosclerotic Plaque in B-mode Carotid Ultrasound Image
Lect. Notes Electrical Eng.
International Conference on Robotics, Control, Automation and Artificial Intelligence ; 2022 November 24, 2022 - November 26, 2022
2023-11-18
13 pages
Aufsatz/Kapitel (Buch)
Elektronische Ressource
Englisch
Photosensitization of Atherosclerotic Plaque
British Library Conference Proceedings | 1994
|BASE | 2017
|Computer Based Analysis of Ultrasound Images for Assessing Carotid Artery Plaque Risk
British Library Conference Proceedings | 2003
|Segmentation of the carotid intima-media region in B-mode ultrasound images
British Library Online Contents | 2010
|