HARNESSING GENERATIVE MODELS FOR MEDICAL IMAGE SYNTHESIS AND AUGMENTATION IN DIAGNOSTIC TASKS
Keywords:
Generative Models, Image Synthesis, Synthetic Data, Medical Imaging, GAN, Healthcare AI, Deep Learning, Data Augmentation, VAE, Diagnostic TasksAbstract
The application of generative models, particularly Generative Adversarial Networks (GANs) and Variational Autoencoders (VAEs), in medical imaging has witnessed rapid evolution. These models hold immense potential for augmenting datasets through synthetic image generation, thereby mitigating issues such as class imbalance and limited data availability. This paper reviews the progress in leveraging generative models for medical image synthesis, focusing on their application in diagnostic imaging tasks. The objective is to understand how synthetic data can enhance deep learning models in terms of performance, generalization, and robustness. It also explores challenges and ethical considerations, offering a structured perspective for future research and clinical integration.
References
Goodfellow, I., Pouget-Abadie, J., Mirza, M., et al. (2014). Generative Adversarial Nets. Advances in Neural Information Processing Systems, 27, 2672–2680.
Frid-Adar, M., Klang, E., Amitai, M., et al. (2018). Synthetic data augmentation using GAN for improved liver lesion classification. Neurocomputing, 321, 321–331. https://doi.org/10.1016/j.neucom.2018.09.013
Shin, H.-C., Tenenholtz, N. A., Rogers, J. K., et al. (2018). Medical image synthesis for data augmentation and anonymization using generative adversarial networks. Medical Image Analysis, 58, 101–110. https://doi.org/10.1016/j.media.2019.101552
Chartsias, A., Joyce, T., Dharmakumar, R., et al. (2017). Adversarial image synthesis for unpaired multi-modal cardiac data. arXiv preprint arXiv:1712.02765
Kingma, D. P., & Welling, M. (2013). Auto-Encoding Variational Bayes. arXiv preprint arXiv:1312.6114
Mahapatra, D., Bozorgtabar, B., Garnavi, R. (2019). Efficient active learning for image classification using a core-set approach. Medical Image Analysis, 60, 101601.
Zhu, J.-Y., Park, T., Isola, P., & Efros, A. A. (2017). Unpaired image-to-image translation using Cycle-Consistent Adversarial Networks. IEEE ICCV, 2223–2232.
Baur, C., Denner, S., Wiestler, B., et al. (2021). Autoencoders for unsupervised anomaly segmentation in brain MR images: A comparative study. Medical Image Analysis, 69, 101952.
Sandfort, V., Yan, K., Pickhardt, P. J., et al. (2019). Data augmentation using GANs for real-time prostate segmentation. Computerized Medical Imaging and Graphics, 77, 101662.
Chaitanya, K., Karani, N., Baumgartner, C. F., et al. (2019). Semi-supervised learning with GANs for cardiac MR segmentation. Medical Image Analysis, 54, 1–7.
