Naslov
Comparison of edge detectors for urinary bladder cancer diagnostic

Autori
Lorencin, Ivan ; Barišić, Borna ; Anđelić, Nikola ; Španjol, Josip ; Car, Zlatan

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

Izvornik
Proceedings of International Conference on Innovative Technologies IN-TECH 2019 / Car, Zlatan ; Kudláček, Jan - Rijeka : Tehnički fakultet Sveučilišta u Rijeci, 2019, 13-16

Skup
International Conference on Innovative Technologies (IN-TECH 2019)

Mjesto i datum
Beograd, Srbija, 11.09.2019. - 13.09.2019

Vrsta sudjelovanja
Predavanje

Vrsta recenzije
Međunarodna recenzija

Ključne riječi
Edge Detector ; Multilayer Perceptron ; Prewitt ; Roberts ; Sobel ; Urinary Bladder Cancer

Sažetak
One of the key challenges in urinary bladder cancer diagnosis is classification of tumor images without need of biopsy and pathohistological analysis. The convolutional neural network (CNN) is commonly used method in medical image classification. In this paper an alternative method to CNN is proposed. In this case, a multilayer perceptron is used alongside Sobel, Prewitt and Roberts edge detectors. By using Sobel edge detector, maximal AUC value of 0.99 is achieved. By using Prewitt and Roberts maximal AUC value of 0.98 is achieved. Results suggest that Prewitt edge detector offers stabile AUC vs. number of hidden layers performances.

Izvorni jezik
Engleski

Znanstvena područja
Elektrotehnika, Računarstvo, Strojarstvo, Kliničke medicinske znanosti

POVEZANOST RADA