METHOD OF GENERATIVE-ADVERSARIAL NETWORKS SEARCHING ARCHITECTURES FOR BIOMEDICAL IMAGES SYNTHESIS
DOI:
https://doi.org/10.15588/1607-3274-2024-1-10Keywords:
generative adversarial network, biomedical images, cytological images, search for neural network architectures, genetic algorithms, FID metrics, computer systems for automatic diagnosticsAbstract
Context. The article examines the problem of automatic design of architectures of generative-adversarial networks. Generativeadversarial networks are used for image synthesis. This is especially true for the synthesis of biomedical images – cytological and histological, which are used to make a diagnosis in oncology. The synthesized images are used to train convolutional neural networks. Convolutional neural networks are currently among the most accurate classifiers of biomedical images.
Objective. The aim of the work is to develop an automatic method for searching for architectures of generative-adversarial networks based on a genetic algorithm.
Method. The developed method consists of the stage of searching for the architecture of the generator with a fixed discriminator and the stage of searching for the architecture of the discriminator with the best generator.
At the first stage, a fixed discriminator architecture is defined and a generator is searched for. Accordingly, after the first step, the architecture of the best generator is obtained, i.e. the model with the lowest FID value.
At the second stage, the best generator architecture was used and a search for the discriminator architecture was carried out. At each cycle of the optimization algorithm, a population of discriminators is created. After the second step, the architecture of the generative-adversarial network is obtained.
Results. Cytological images of breast cancer on the Zenodo platform were used to conduct the experiments. As a result of the study, an automatic method for searching for architectures of generatively adversarial networks has been developed. On the basis of computer experiments, the architecture of a generative adversarial network for the synthesis of cytological images was obtained. The total time of the experiment was ~39.5 GPU hours. As a result, 16,000 images were synthesized (4000 for each class). To assess the quality of synthesized images, the FID metric was used.The results of the experiments showed that the developed architecture is the best. The network’s FID value is 3.39. This result is the best compared to well-known generative adversarial networks.
Conclusions. The article develops a method for searching for architectures of generative-adversarial networks for the problems of synthesis of biomedical images. In addition, a software module for the synthesis of biomedical images has been developed, which can be used to train CNN.
References
Goodfellow I., Pouget-Abadie J., Mirza M. et al. Generative Adversarial Nets, Advances in Neural Information Processing Systems, 2014, №27.
Salimans T., Goodfellow I., Zaremba W. et al. Improved Techniques for Training GANs [Electronic resource], arXiv.org, 2016. Access mode: https://arxiv.org/abs/1606.03498.
Karras T., Aila T., Laine S. et al. Progressive Growing of GANs for Improved Quality, Stability, and Variation [Electronic resource], arXiv.org, 2017. Access mode: https://arxiv.org/abs/1710.10196.
Analyzing and Improving the Image Quality of StyleGAN [Electronic resource], Conference on Computer Vision and Pattern Recognition (CVPR) : proceedings, Seattle, 2020,P. 8107–8116. Access mode: https://doi.org/10.1109/CVPR42600.2020.00813.
Arjovsky M., Chintala S. Wasserstein Generative Adversarial Networks [Electronic resource], International Conference on Machine Learning : proceedings, 2017. Access mode: https://leon.bottou.org/publications/pdf/icml2017.pdf.
Düzyel O., Kuntalp M. Data Augmentation with GAN increases the Perfor-mance of Arrhythmia Classification for an Unbalanced Dataset [Electronic resource], arXiv.org. – 2023. Access mode: https://doi.org/10.48550/arXiv.2302.13855.
Radford A., Metz L. Unsupervised Representation Learning with Deep Convolutional Generative Adversarial Networks [Electronic resource], arXiv.org, 2016. Access mode: https://doi.org/10.48550/arXiv.2302.13855.
Costa Pedro et al. End-to-End Adversarial Retinal Image Synthesis [Electronic resource], IEEE Transactions on Medical Imaging, 2018, Vol. 37, № 3, pp. 781–791. Access mode: https://doi.org/10.1109/tmi.2017.2759102.
Frid-Adar Maayan et al. GAN-based synthetic medical image augmentation for increased CNN performance in liver lesion classification [Electronic resource], Neurocomputing. 2018, Vol. 321, pp. 321–331. Access mode: https://doi.org/10.1016/j.neucom.2018.09.013.
Zoph B., Le Q. Neural Architecture Search with Reinforcement Learning [Electronic resource], arXiv.org, 2017. Access mode: https://doi.org/10.48550/arXiv.1611.01578.
Gong Xinyu et al. AutoGAN: Neural Architecture Search for Generative Adversarial Networks [Electronic resource], 2019 IEEE/CVF International Conference on Computer Vision (ICCV). Seoul, Korea (South), 2019. Access mode: https://doi.org/10.1109/iccv.2019.00332.
Baker B., Gupta O. Designing Neural Network Architectures using Rein-forcement Learning [Electronic resource], arXiv.org, 2017. Access mode: https://doi.org/10.48550/arXiv.1611.02167.
Ying G., He X., Gao B. et al. Efficient Two-stage Evolutionary Architecture Search for GANs [Electronic resource], arXiv.org, 2020. Access mode: https://doi.org/10.48550/arXiv.2111.15097.
Gao Chen et al. AdversarialNAS: Adversarial Neural Architecture Search for GANs [Electronic resource], 2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR). Seattle, WA, USA, 2020. Access mode: https://doi.org/10.1109/cvpr42600.2020.00572.
Brock A., Donahue J., Simonyan K. Large Scale GAN Training for High Fidelity Natural Image Synthesis [Electronic resource], Semantic Schoola, 2018. Access mode: https://www.semanticscholar.org/paper/Large-Scale-GANTraining-for-High-Fidelity-Natural-BrockDonahue/22aab110058ebbd198edb1f1e7b4f69fb13c0613.
Berezsky O. M., Liashchynskyi P. B., Pitsun O. Y. et al. Deep network-based method and software for small sample biomedical image generation and classification, Radio Electronics, Computer Science, Control, 2024, № 4, P. 76.
He Kaiming et al. Deep Residual Learning for Image Recognition [Electronic resource], 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR). Las Vegas, NV, USA, 2016. Access mode: https://doi.org/10.1109/cvpr.2016.90.
Vaswani A., Shazeer N., Parmar N. et al. Attention is All you Need [Electronic resource], Semantic Schoolar, 2017. Access mode: https://www.semanticscholar.org/paper/Attention-is-Allyou-Need-Vaswani-Shazeer/204e3073870fae3d05bcbc2f6a8e263d9b72e776.
Sunkara R., Luo T. No More Strided Convolutions or Pooling: A New CNN Building Block for Low-Resolution Images and Small Objects [Electronic resource], Machine Learning and Knowledge Discovery in Databases. Cham, 2023, pp. 443–459. Access mode: https://doi.org/10.1007/978-3-031-26409-2_27.
Barratt S., Sharma R. A Note on the Inception Score. [Electronic resource], arXiv.org, 2018. Access mode: https://arxiv.org/abs/1801.01973.
Borji A. Pros and cons of GAN evaluation measures [Electronic resource], Computer Vision and Image Understanding, 2019, Vol. 179, pp. 41–65. Access mode: https://doi.org/10.1016/j.cviu.2018.10.009.
Zhao S., Liu Z., Lin J. et al. Differentiable Augmentation for Data-Efficient GAN Training [Electronic resource], arXiv.org, 2020. Access mode: https://arxiv.org/abs/2006.10738.
Real Esteban et al. Regularized Evolution for Image Classifier Architecture Search [Electronic resource], Proceedings of the AAAI Conference on Artificial Intelligence, 2019, Vol. 33, pp. 4780–4789. Access mode: https://doi.org/10.1609/aaai.v33i01.33014780.
Cytological and histological images of breast cancer. (n.d.) [Electronic resource] // Zenodo. – 2023. – Access mode: https://doi.org/10.5281/zenodo.7890874.
Liashchynskyi P. Rudi – Python library [Electronic resource], GitHub, 2018. Access mode: https://github.com/liashchynskyi/rudi.
Ye J., Lim J. Geometric GAN [Electronic resource], arXiv.org, 2017. – Access mode: https://arxiv.org/abs/1705.02894.
Kingma D., Ba J. Adam: A Method for Stochastic Optimization [Electronic resource], arXiv.org, 2017. Access mode: https://arxiv.org/abs/1412.6980.
Bynagari N. B. GANs Trained by a Two Time-Scale Update Rule Converge to a Local Nash Equilibrium [Electronic resource], Asian Journal of Applied Science and Engineering. 2019, Vol. 8, № 1, pp. 25–34. Access mode: https://doi.org/10.18034/ajase.v8i1.9.
Miyato T., Kataoka T., Koyama M. Spectral Normalization for Generative Adversarial Networks [Electronic resource], arXiv.org, 2018. Access mode: https://arxiv.org/abs/1802.05957.
Berezsky O. M. Liashchynskyi P. B. Comparison of generative adversarial networks architectures for biomedical images synthesis [Electronic resource], Applied Aspects of Information Technology, 2021, Vol. 4, № 3, pp. 250–260. Access mode: https://doi.org/10.15276/aait.03.2021.4.
Berezsky О. М., Berezska К. M., Melnyk G. M. et al. Design of computer systems for biomedical image analysis, Proceedings of the X th International Conference «The Experience of Designing and Application of CAD Systems in Microelectronics» CADSM 2009. Lviv, Publishing House Vezha&CoС, pp. 186–191.
Berezsky O. Verbovyy S., Datsko T. The intelligent system for diagnosing breast cancers based on image analysis [Electronic resource], 2015 Information Technologies in Innovation Business Conference (ITIB). Kharkiv, Ukraine, 7–9 October 2015, 2015. Access mode: https://doi.org/10.1109/itib.2015.7355067.
Berezsky Oleh et al. Fuzzy system diagnosing of precancerous and cancerous conditions of the breast [Electronic resource, 2016 XIth International Scientific and Technical Conference “Computer Sciences and Information Technologies (CSIT). Lviv, Ukraine, 6–10 September 2016. 2016. Access mode: https://doi.org/10.1109/stc-csit.2016.7589906.
Berezsky O., Verbovyy S., Pitsun O. Hybrid Intelligent Information Technology for Biomedical Image Processing [Electronic resource], IEEE 13th International Scientific and Technical Conference on Computer Sciences and Information Technologies, 2018. Access mode: https://ieeexplore.ieee.org/document/8526711.
Berezsky O., Dubchak L., Batryn T.et al. Fuzzy System For Breast Disease Diagnosing Based On Image Analysis [Electronic resource], Proceedings of the II International Workshop Informatics & Data-Driven Medicine (IDDM 2019). – 2019. – Access mode: http://ceur-ws.org/Vol2488/paper6.pdf.
Downloads
Published
How to Cite
Issue
Section
License
Copyright (c) 2024 O. M. Berezsky, P. B. Liashchynskyi
This work is licensed under a Creative Commons Attribution-ShareAlike 4.0 International License.
Creative Commons Licensing Notifications in the Copyright Notices
The journal allows the authors to hold the copyright without restrictions and to retain publishing rights without restrictions.
The journal allows readers to read, download, copy, distribute, print, search, or link to the full texts of its articles.
The journal allows to reuse and remixing of its content, in accordance with a Creative Commons license СС BY -SA.
Authors who publish with this journal agree to the following terms:
-
Authors retain copyright and grant the journal right of first publication with the work simultaneously licensed under a Creative Commons Attribution License CC BY-SA that allows others to share the work with an acknowledgement of the work's authorship and initial publication in this journal.
-
Authors are able to enter into separate, additional contractual arrangements for the non-exclusive distribution of the journal's published version of the work (e.g., post it to an institutional repository or publish it in a book), with an acknowledgement of its initial publication in this journal.
-
Authors are permitted and encouraged to post their work online (e.g., in institutional repositories or on their website) prior to and during the submission process, as it can lead to productive exchanges, as well as earlier and greater citation of published work.
