20.500.12556/RUL-185989 Zasnova radiologom uporabnih razlag napovedi raka dojk s segmentacijo mamografskih slik Design of Explanations of Breast Cancer Predictions for Radiologists Using Mammographic Segmentation Rak dojk letno prizadene približno 2,3 milijona žensk po svetu in v Sloveniji povzroči več kot 1.300 novih diagnoz. Presejalne teste kljub učinkovitosti omejujejo velika količina podatkov, variabilnost med ocenjevalci ter lažni rezultati. Čeprav obstajajo metode umetne inteligence za napovedovanje, te pogosto ne vzpostavljajo zadostnega zaupanja pri radiologih. V tem magistrskem delu je predstavljena zasnova radiologom uporabnih napovedi raka dojk, ki temelji na segmentaciji mamografskih slik. Po začetni primerjavi treh modelov (UNeXt, SegFormer in MedSAM) smo izbrali prednaučen model MedSAM, ki smo ga doučili z uporabo podatkovnega nabora CSAW-CC. Napovedi so bile izvedene na izsekih slik s kasnejšo rekonstrukcijo celotnih mamografskih slik, pri čemer so lezije na končni sliki jasno predstavljene z obkrožitvijo. Na testni množici sta bila dosežena koeficient Dice 0,72 in IoU 0,59 – kar je primerljivo z literaturo. V anketi so radiologi potrdili zadovoljstvo z napovedmi. Delo prispeva k zasnovi in razvoju razložljivih orodij, ki so primarno namenjena končnim uporabnikom – radiologom. Breast cancer annually affects approximately 2.3 million women worldwide and results in more than 1,300 new diagnoses in Slovenia. Despite their effectiveness, screening tests are limited by the large volume of data, interobserver variability, and false results. Although artificial intelligence methods for prediction exist, they often fail to establish sufficient trust among radiologists. This master’s thesis presents the design of explanations of breast cancer predictions for radiologists based on the segmentation of mammographic images. After an initial comparison of three models (UNeXt, SegFormer, and MedSAM), we selected the pre-trained MedSAM model, which we fine-tuned using the CSAW-CC dataset. Predictions were performed on image patches, followed by reconstruction of the full mammographic images, with lesions on the final image clearly represented by outlining. On the test set, a Dice coefficient of 0.72 and an IoU of 0.59 were achieved, which is comparable to the literature. In the survey, radiologists confirmed their satisfaction with the predictions. The work contributes to the design and development of explainable tools that are primarily intended for end users – radiologists. razložljiva umetna inteligenca globoko učenje segmentacija mamografske slike rak dojk explainable artificial intelligence deep learning segmentation mammograms breast cancer true false false Slovenski jezik Angleški jezik Magistrsko delo/naloga 2026-08-25 10:30:09 2026-08-25 10:30:15 2026-08-28 07:52:12 0000-00-00 00:00:00 2026 0 0 0000-00-00 NiDoloceno NiDoloceno NiDoloceno 0000-00-00 0000-00-00 0000-00-00 38489 289189379 Rupnik_Lenart_-_Zasnova_radiologom_uporabnih_razlag_napovedi_raka_dojk_s_segmentacijo_mamografsk.pdf Rupnik_Lenart_-_Zasnova_radiologom_uporabnih_razlag_napovedi_raka_dojk_s_segmentacijo_mamografsk.pdf 1 354626FED560FF36D4DD1BC33FCE0D13 48670f295f8754c77fc8517e553fc6cf16977d7356928113d194f8dfd328117e cb5e8057-a05e-11f1-9b0d-0050569b8976 https://repozitorij.uni-lj.si/Dokument.php?lang=slv&id=245027 Fakulteta za računalništvo in informatiko 0 0 0