Fake Data Generation for Medical Image Augmentation using GANs

This paper uses WGAN-GP to generate fake data that can be used as augmented data for strabismus classification and analyze the results. In the introduction of this paper, the general diagnostic technique for strabismus disease is described and the diagnostic technique using deep learning is describe...

Full description

Saved in:
Bibliographic Details
Published in2022 International Conference on Artificial Intelligence in Information and Communication (ICAIIC) pp. 197 - 199
Main Authors Kim, Donghwan, Joo, Jaehan, Kim, Suk Chan
Format Conference Proceeding
LanguageEnglish
Published IEEE 21.02.2022
Subjects
Deep learning
diagnosis
Diseases
Learning (artificial intelligence)
Measurement
Medical diagnostic imaging
strabismus
WGAN-GP
Online AccessGet full text

Cover

More Information
Summary:This paper uses WGAN-GP to generate fake data that can be used as augmented data for strabismus classification and analyze the results. In the introduction of this paper, the general diagnostic technique for strabismus disease is described and the diagnostic technique using deep learning is described. And the reason for generating fake data is described. Main subject describes the WGAN-GP, data set used for data generation and evaluation metrics of GAN. In the experimental result, the data generated by the GAN is visually checked, and the performance of the fake data is evaluated with the FID that is one of the evaluation metrics of the GAN. And in the conclusion, evaluation of the proposed GAN and future work are described.
DOI:10.1109/ICAIIC54071.2022.9722700