CONFLUX: A Latent Diusion Model for 3D Chest-CT Synthesis with RL Post-Training

Researchers developed CONFLUX, a 3D latent diffusion model for chest CT generation that achieves high-fidelity results and enables control over clinical attributes. The model uses reinforcement learning post-training to improve control over requested clinical findings, reducing the gap to real-scan reliability by 47%.

1sec.ai