This repository accompanies the research paper “LungGen: Diffusion-Based Synthetic Chest X-ray Generation” and provides supplementary materials supporting the study. The upload includes: The full research paper describing a DDPM-based framework for synthetic chest X-ray generation. Generated synthetic chest X-ray samples at 256×256 resolution produced by the LungGen model. Representative original samples from the NIH Chest X-ray dataset. Corresponding preprocessed versions of the NIH images used during model training. A demonstration video visualizing the progressive denoising process of the DDPM model, illustrating how image quality improves across training epochs. The synthetic images aim to capture anatomically plausible thoracic structures while preserving patient privacy. These materials are intended for academic, educational, and research purposes, enabling transparency, reproducibility, and qualitative assessment of diffusion-based medical image generation. No patient-identifiable information is included. All original dataset samples are shared in accordance with their respective usage and citation policies. This repository supports qualitative evaluation and methodological transparency for diffusion-based medical image synthesis.
P et al. (Mon,) studied this question.