GENETIC ALGORITHM BASED ATTENTION UNET OPTIMIZATION FOR BREAST TUMOR SEGMENTATION.

OPTIMIZACIJA UNET MODELA ZA SEGMENTACIJU TUMORA DOJKE ZASNOVANOG NA GENETIČKOM ALGORITAMU.

WORLD Health Organization ; INTERNATIONAL Agency for Research on Cancer

As one of the main causes of cancer-related mortality among women worldwide, breast cancer requires better diagnostic techniques that can provide non-invasive, fast, and accurate detection. The World Health Organization (WHO) has a dedicated cancer agency called the International Agency for Research on Cancer (IARC), whose mission is to undertake and coordinate research on cancer causes. Mammography is one of many imaging modalities that is frequently used to find abnormalities. Although automated breast mass segmentation in mammography is vital, the uniform sizes and shapes of tumors make it a difficult process. UNet models have shown a significant segmentation in the medical images. In this study, we propose a prominent genetic algorithm (GA) for the generation of UNet models by selecting the optimal parameters. The experiments involved manually generated architectures, basic UNet model and an attention based UNet, AUNet model with different filter sizes. As a result of the manual approach, the AUNet outperformed the base model and hence the AUNet is considered as the base model for the GA process. The experiments show that the models evolved using GA are simple and are of small architecture. The model yielded a better segmentation of the images and outperformed the manually created UNet models, with dice scores and Intersection over Union (IoU) scores of 91.6% and 89.2%, respectively. [ABSTRACT FROM AUTHOR]

Kao jedan od glavnih uzroka smrtnosti od raka kod žena širom sveta, rak dojke zahteva bolje dijagnostičke tehnike koje mogu da obezbede neinvazivno, brzo i tačno otkrivanje. Svetska zdravstvena organizacija (SZO) ima namensku agenciju za rak pod nazivom Međunarodna agencija za istraživanje raka (IARC), čija je misija da preduzima i koordinira istraživanje o uzrocima raka. Mamografija je jedan od mnogih modaliteta snimanja koji se često koristi za otkrivanje abnormalnosti. Iako je automatska segmentacija mase dojke u mamografiji od vitalnog značaja, ujednačene veličine i oblici tumora čine to teškim procesom. UNet modeli su pokazali značajnu segmentaciju na medicinskim slikama. U ovoj studiji predlažemo istaknuti genetski algoritam (GA) za generisanje UNet modela izborom optimalnih parametara. Eksperimenti su uključivali ručno generisane arhitekture, osnovni UNet model i UNet, AUNet model zasnovan na različitim veličinama filtera. Kao rezultat ručnog pristupa, AUNet je nadmašio osnovni model i stoga se AUNet smatra osnovnim modelom za GA proces. Eksperimenti pokazuju da su modeli razvijeni korišc'enjem GA jednostavni i male arhitekture. Model je dao bolju segmentaciju slika i nadmašio je ručno kreirane UNet modele, sa rezultatima kockice i Intersection over Union (IoU) rezultatima od 91,6% i 89,2%, respektivno. [ABSTRACT FROM AUTHOR]