# Implementation of a Conditional Latent Diffusion-Based Generative Model to Synthetically Create Unlabeled Histopathological Images

## TL;DR

This paper explores using a conditional latent diffusion model to generate synthetic histopathological images from unlabeled data, improving controllability and quality.

## Contribution

The novel approach combines VQ-GAN, diffusion models, and clustering with expert input for conditional histopath image generation.

## Key findings

- Clustering latent features improved the controllability of synthetic image generation.
- Expert input enhanced the interpretability of the clustering process.
- Quantitative metrics confirmed the high quality of the generated histopathological images.

## Abstract

Generative image models have revolutionized artificial intelligence by enabling the synthesis of high-quality, realistic images. These models utilize deep learning techniques to learn complex data distributions and generate novel images that closely resemble the training dataset. Recent advancements, particularly in diffusion models, have led to remarkable improvements in image fidelity, diversity, and controllability. In this work, we investigate the application of a conditional latent diffusion model in the healthcare domain. Specifically, we trained a latent diffusion model using unlabeled histopathology images. Initially, these images were embedded into a lower-dimensional latent space using a Vector Quantized Generative Adversarial Network (VQ-GAN). Subsequently, a diffusion process was applied within this latent space, and clustering was performed on the resulting latent features. The clustering results were then used as a conditioning mechanism for the diffusion model, enabling conditional image generation. Finally, we determined the optimal number of clusters using cluster validation metrics and assessed the quality of the synthetic images through quantitative methods. To enhance the interpretability of the synthetic image generation process, expert input was incorporated into the cluster assignments.