의료영상 분류를 위한 심층신경망 훈련에서 StyleGAN 합성 영상의 데이터 증강 효과 분석
Data Augmentation Effect of StyleGAN-Generated Images in Deep Neural Network Training for Medical Image Classification
1
조회
0
다운로드
0
인용
0
참고문헌
키워드
초록
본 논문에서는 의료 영상 분류를 위한 심층 신경망 훈련에서 StyleGAN 합성 영상의 데이터 증강 효과를 분석한다. 이를 위해흉부 X선 영상에서의 폐렴 진단과 복부 CT 영상에서의 간전이암 분류 문제에서 StyleGAN 합성 영상을이용하여 VGG-16 심층 합성곱 신경망 훈련을 수행한다. 실험에서 분류 결과에 대한 정량적, 정성적 분석을 통해 StyleGAN 데이터 증강이 특징공간에서 클래스 외곽을 확장하는 특성을 보이며, 이와 같은 특성으로 인해 실제 영상과의 적절한 비율을 통해 혼합했을 때분류 성능이 개선될 수 있음을 확인하였다.
Abstract
In this paper, we examine the effectiveness of StyleGAN-generated images for data augmentation in training deep neural networks for medical image classification. We apply StyleGAN data augmentation to train VGG-16 networks for pneumonia diagnosis from chest X-ray images and focal liver lesion classification from abdominal CT images. Through quantitative and qualitative analyses, our experiments reveal that StyleGAN data augmentation expands the outer class boundaries in the feature space. Thanks to this expansion characteristics, the StyleGAN data augmentation can enhance classification performance when properly combined with real training images.
참고문헌 (20)
- J. Garstka, "Pneumonia detection in x-ray chest images based on convolutional neural networks and data augmentation methods", 2020 Signal Processing: Algorithms, Architectures, Arrangements, and Applications, pp.18~23 (2020), Institute of Electrical and Electronics Engineers
- S. Motamed, "Data augmentation using generative adversarial networks(gans)for gan-based detection of pneumonia and covid-19 in chestx-ray images", Informatics in Medicine Unlocked, 27, pp.100779~ (2021)
- A. Radford, "Unsupervised rep-resentation learning with deep convolutional generativead-versarial networks", International Conference on Learning Representations (2016)
- T. Karras, "Progressive growing of gans for improved quality, stability, andvaria-tion", inInternational Conferenceon Learnin gRepresenta-tions (2017)
- T. Karras, "Astyle-basedgeneratorar-chitecture for generative adversarial networks", Compute rVision and Pattern Recognition (2019)
- H. Salehinejad, "Generalization of deep neural networks for chest pathol-ogy classification in x-rays using generative adversarial net-works", Institute of Electricaland Electronics Engineerin-ternational conference on acoustics, speech and signal pro-cessing (2018)
- 김덕선, "복부ct영상에서 심층 합성곱 신경망 기반의 국소 간 병변 분류를 위한 데이터 증강 기법의 효과", 한국컴퓨터그래픽스학회논문지, 29(2), pp.1~11 (2023)
- D. Zhao, "Syn-thetic medical images using f&bgan for improvedl ungnod-ules classification by multi-scale vgg-16", Symmetry, 10(10), pp.519~ (2018)
- H. Lee, "Classi-fication of focal liver lesions in ct images using convolutional neural networks with lesion information augmented patches and synthetic data augmentation", Medical Physics, 48(9), pp.5029~5046 (2021)
- A. Woo, "Classification of focal liver lesions in abdominal ctimages using convolutional neu-ral networks with stylegan data augmentation", Proceed-ing of the Fall Conference of the Korea Multimedia Society, 26(1), pp.94~96 (2023)
- M. Loey, "Within the lack of chest covid-19x-ray dataset : a novel detection model based on gan and deep transfer learning", Symmetry, 12(4), pp.651~ (2020)
- S. Buragadda, "Hcugan : Hybrid cyclic unet gan for generating augmented synthetici mages of chestx-ray images for multi classifica-tion of lung diseases", International Journal of Engineering Trends and Technology, 70(2), pp.229~238 (2022)
- L. vander Maaten, "Visualizing data usingt-sne", Journal of Machine Learning Research, 9, pp.2579~2605 (2008)
- R. Hao, "A comprehensive study of data augmentation strategies for prostate cancer detection in diffusion-weighted mri using convolutional neural networks", Journal of Digital Imaging, pp.862~876 (2021)
- M. Kim, "Data augmentation techniques for deeplearning based medical image analyses", Journal of the Ko-rean Society of Radiology, 81(6) (2020)
- M. Nishio, "Auto-matic classification between covid-19 pneumonia, non-covid-19 pneumonia, and the healthy on chestx-ray image : com-bination of data augmentation methods", Scientific Reports, 10(1), pp.1~6 (2020)
- D. Kermany, "Identifying medical diagnoses and treatable diseases by image-based deep learning", Cell, 172(5), pp.1122~1131 (2018)
- K. Asnaoui, "Artificial intelligence and blockchain for future cybersecurity applications", pp.257~284 (2021), Springer International Publishing
- I. Rudan, "Epidemiology and etiology of childhood pneumonia", Bulletin of theWorld Health Organization, 86, pp.408B~416B (2008)
- H. Zar, "Advances in the diagnosis of pneumonia in children", British Medical Journal, 358 (2017)