Here is the article that Dr. Umer and I will be building on (it's also the paper that I initially reached out to talk about):
CytoDiff: https://arxiv.org/abs/2507.05063v2
Very similar articles we're building upon can be found below:
CytoSyn: https://arxiv.org/abs/2603.18089
This paper is about a foundational model approach for pathology and histopathology. They utilize embeddings and different adaptation techniques to show how larger pretrained models can be utilized for specific medical imaging tasks more effectively.
MorphoDiff: https://www.biorxiv.org/content/10.1101/2024.12.19.629451v1
This paper uses a StableDiffusion model similar to the one we would be employing. However, they focus more on perturbation modeling and cellular morphology. This research looks at how synthetic variations of tissue help downstream tasks such as drug discovery. While more adjacent to my topic, I still find it fascinating for the potential uses for drug discovery.
PathDiff: https://arxiv.org/abs/2506.23440
This paper explores a diffusion-based generation technique specifically for histopathology images. They focus on improving image quality and having a diverse range of images while preserving the biologically meaningful features. This paper furthers the potential for diffusion models to expand medical imaging datasets and helps support future pathology research.
Unleashing the Potential of Synthetic Images: A Study on Histopathology Image Classification: https://arxiv.org/abs/2409.16002?utm_
This study examines how synthetic images can help doctors and researchers train better disease-detection systems in limited data environments. It studies whether synthetic microscope images of tissue samples can improve AI's ability to detect cancer.