Pregled bibliografske jedinice broj: 1139233
Automatic extraction of multiple-study X-ray images
Automatic extraction of multiple-study X-ray images // 2021 International Conference on INnovations in Intelligent SysTems and Applications (INISTA)
Köseköy, Turska, 2021. 87, 6 doi:10.1109/INISTA52262.2021.9548551 (predavanje, međunarodna recenzija, cjeloviti rad (in extenso), znanstveni)
CROSBI ID: 1139233 Za ispravke kontaktirajte CROSBI podršku putem web obrasca
Naslov
Automatic extraction of multiple-study X-ray images
Autori
Dumenčić, Stella ; Tschauner, Sebastian ; Hržić, Franko ; Štajduhar, Ivan
Vrsta, podvrsta i kategorija rada
Radovi u zbornicima skupova, cjeloviti rad (in extenso), znanstveni
Izvornik
2021 International Conference on INnovations in Intelligent SysTems and Applications (INISTA)
/ - , 2021
ISBN
978-1-6654-3603-8
Skup
International Conference on INnovations in Intelligent SysTems and Applications (INISTA)
Mjesto i datum
Köseköy, Turska, 25.08.2021. - 27.08.2021
Vrsta sudjelovanja
Predavanje
Vrsta recenzije
Međunarodna recenzija
Ključne riječi
object detection ; YOLO ; radiograph ; image segmentation
Sažetak
In medical radiology standard practice, radiographs of different projections or studies are occasionally merged into a single image for convenience. Before the radiographs can be utilized for further automatic processing, e.g., for detecting and/or localizing specific injuries, such merged projections or studies need to be separated. Doing this manually can be tiresome ; therefore, we decided to investigate the possibility of automating this process using contemporary image processing and machine learning techniques. We implemented two independent automatic solutions for this problem: (1) a manually-tailored image-splitting method using morphological operations, image binarization, and boundary-box detection, and (2) an instance of the YOLOv4 algorithm. The manually-tailored algorithm and YOLOv4 model were trained on the training data set of 4, 000 images and validated on the data set of 250 images. Comparing both on a disjoint test set, consisting of 250 images, the YOLOv4 model noticeably outperformed the manually-tailored method with an accuracy of 0.992, F1 score of 0.996, and the intersection-overunion of 0.8978±0.079. The results suggest that YOLOv4 can efficiently extract large portions of embedded projections or studies from medical radiographs.
Izvorni jezik
Engleski
Znanstvena područja
Računarstvo, Kliničke medicinske znanosti
POVEZANOST RADA
Projekti:
HRZZ-IP-2020-02-3770 - Strojno učenje za prijenos znanja u medicinskoj radiologiji (RadiologyNET) (Štajduhar, Ivan, HRZZ - 2020-02) ( CroRIS)
NadSve-Sveučilište u Rijeci-uniri-tehnic-18-15 - Razvoj postupaka temeljenih na strojnom učenju za prepoznavanje bolesti i ozljeda iz medicinskih slika (Štajduhar, Ivan, NadSve ) ( CroRIS)
Ustanove:
Tehnički fakultet, Rijeka