# Unsupervised Histopathology Image Synthesis

arXiv: https://arxiv.org/abs/1712.05021 · HTML: https://mlsanity.com/paper/1712.05021/

## Key facts

- Cited by 11 papers in the map (data through 7 Oct 2026).
- First posted on arXiv 13 Dec 2017.

## Built on by (most cited first, 11 of 11)

- [Ensembling with Deep Generative Views](https://mlsanity.com/paper/2104.14551/) (2021)
- [DiffInfinite: Large Mask-Image Synthesis via Parallel Random Patch Diffusion in Histopathology](https://mlsanity.com/paper/2306.13384/) (2023)
- [Random Multi-Channel Image Synthesis for Multiplexed Immunofluorescence Imaging](https://mlsanity.com/paper/2109.09004/) (2021)
- [An expert-driven data generation pipeline for histological images](https://mlsanity.com/paper/2406.01403/) (2024)
- [Improving text-conditioned latent diffusion for cancer pathology](https://mlsanity.com/paper/2412.06487/) (2024)
- [Synthetic Generation of Three-Dimensional Cancer Cell Models from Histopathological Images](https://mlsanity.com/paper/2101.11600/) (2021)
- [Neuroplastic graph attention networks for nuclei segmentation in histopathology images](https://mlsanity.com/paper/2201.03669/) (2022)
- [Preparing data for pathological artificial intelligence with clinical-grade performance](https://mlsanity.com/paper/2205.10748/) (2022)
- [Leveraging weak complementary labels to improve semantic segmentation of hepatocellular carcinoma and cholangiocarcinoma in H&E-stained slides](https://mlsanity.com/paper/2302.01813/) (2023)
- [Microscopy Image Segmentation via Point and Shape Regularized Data Synthesis](https://mlsanity.com/paper/2308.09835/) (2023)
- [Method and Software Tool for Generating Artificial Databases of Biomedical Images Based on Deep Neural Networks](https://mlsanity.com/paper/2405.16119/) (2024)

## Abstract (arXiv)

Hematoxylin and Eosin stained histopathology image analysis is essential for the diagnosis and study of complicated diseases such as cancer. Existing state-of-the-art approaches demand extensive amount of supervised training data from trained pathologists. In this work we synthesize in an unsupervised manner, large histopathology image datasets, suitable for supervised training tasks. We propose a unified pipeline that: a) generates a set of initial synthetic histopathology images with paired information about the nuclei such as segmentation masks; b) refines the initial synthetic images through a Generative Adversarial Network (GAN) to reference styles; c) trains a task-specific CNN and boosts the performance of the task-specific CNN with on-the-fly generated adversarial examples. Our main contribution is that the synthetic images are not only realistic, but also representative (in reference styles) and relatively challenging for training task-specific CNNs. We test our method for nucleus segmentation using images from four cancer types. When no supervised data exists for a cancer type, our method without supervision cost significantly outperforms supervised methods which perform across-cancer generalization. Even when supervised data exists for all cancer types, our approach without supervision cost performs better than supervised methods.