engleski

Comparison of Image Augmentation Techniques in Urinary Bladder Cancer Diagnostics

Urinary bladder cancer is one of the most common malignancies in the urinary tract. One of the key challenges in medical image analysis is the curation of sufficiently large data sets. For these reasons, in this paper, two different approaches for image data set augmentation are presented. One approach is based on a pipeline that consists of various geometrical variations of original images. The second approach is based on the utilization of Generative Adversarial Networks (GAN). Such an approach is used to artificially create new images that are used to increase the volume of the training data set. The performances were investigated by using urinary bladder cancer data set. From the achieved results, it can be concluded that by applying image augmentation techniques, significantly higher results are achieved. Furthermore, it can be concluded that by applying GAN-based augmentation, even higher results ara achieved

CNN, data set augmentation, GAN, urinary bladder cancer

nije evidentirano