New 3D CT-to-PET translation uses latent diffusion models

Researchers have developed a novel 3D CT-to-PET translation framework using latent Brownian Bridge Diffusion (BBDM). This two-stage method first employs a Variational Autoencoder (VAE) with contrastive learning to align latent representations of anatomical (CT) and metabolic (PET) data. Subsequently, a BBDM translates these aligned latent representations, enabling the synthesis of PET-like information from CT scans. The approach aims to reduce radiation exposure and costs associated with PET imaging, showing improved performance in preserving metabolic activity and lesion detection compared to existing methods. AI

IMPACT This research could lead to reduced radiation exposure and costs in cancer diagnosis by enabling virtual PET imaging from CT scans.

RANK_REASON Academic paper detailing a new method for medical image translation. [lever_c_demoted from research: ic=1 ai=1.0]

Read on arXiv cs.CV →

• computed tomography
• Francesco Di Feola
• Latent Brownian Bridge Diffusion
• positron emission tomography
• variational auto-encoder

• paper
• model release

AI-generated summary · Google Gemini · from 1 sources. How we write summaries →