Summary
This study applies Diffusion Models to synthesize chest X-ray images from the NIH Chest X-rays and Chest X-Ray Images datasets for medical diagnostic data augmentation. The model is fine-tuned conditioned on pathology labels to generate disease-specific X-rays. Evaluation focuses quantitatively on the fidelity, sharpness, diversity, and clinical realism of the generated images.
Generated Samples
Comparison Real-Synthetic Chest X-Ray 1
Comparison Real-Synthetic Chest X-Ray 2
Conditioned Synthetic Sample 3
Project Information
This project was developed as the final project of the PythonML Summer Course, focusing on Chest X-Ray Synthetic Image Generation Using Diffusion Models on Cloud for Disease Diagnosis.
The project was conducted under the HCMUT EE Machine Learning & IoT Lab as part of the summer course activities in 2026.