Naslov
Deep learning methods for detection of carotid artery wall

Autori
Miloš Anić, Branko Arsić, Smiljana Đorović, Nenad Filipović

Vrsta, podvrsta i kategorija rada
Radovi u zbornicima skupova, cjeloviti rad (in extenso), znanstveni

Izvornik
Ri-STEM-2021 Proceedings / Lorencin, Ivan ; Baressi Šegota, Sandi ; Car, Zlatan - Rijeka, 2021, 135-139

ISBN
978-953-8246-22-7

Skup
International Student Scientific Conference (Ri-STEM 2021)

Mjesto i datum
Rijeka, Hrvatska, 10.06.2021. - 11.06.2021

Vrsta sudjelovanja
Predavanje

Vrsta recenzije
Međunarodna recenzija

Ključne riječi
Carotid artery, deep neural networks, U-Net, segmentation, medical images, stenosis, SegNet

Sažetak
Carotid artery is the main artery located in human neck. Its main role is to deliver blood to the neck and face muscles as well as, most importantly, to the brain. Carotid artery stenosis is one of many fatal carotid artery diseases involving carotid artery. Development of stenosis on artery wall can cause brain stroke if plaque breaks. Convolutional neural networks (CNNs) proved to be successful in object classification on images as well as object detection on same images. In the field of segmentation of clinical images, U-Net and SegNet architectures proved to have good performances. The aim of this paper was to use CNN to detect carotid artery wall in order to separate artery tissue from stenosis. Automatic segmentation of carotid artery wall was done via SegNet CNN and was compared with modified U-Net based deep convolutional network. Proposed model was evaluated on the images of real patients which were acquired through ultrasound. Experimental results show that this model outperforms models of other deep neural networks.

Izvorni jezik
Engleski

Znanstvena područja
Računarstvo, Kliničke medicinske znanosti, Biotehnologija u biomedicini (prirodno područje, biomedicina i zdravstvo, biotehničko područje)

POVEZANOST RADA