Verified, then logged.
Sun 22 Nov 2026 · 00:21 KST / 15:21 UTC한국어 | ENSearch /

Breaking01 / AI

LUCID uses chest CT pretraining to correct cardiac PET scans

A clinician seen from behind reviews patient laboratory results alongside reference material and an unfinished AI recommendation at a hospital workstation, with all screen content unreadable. AI-generated image

Researchers proposed LUCID, a model pretrained on 10,000 chest CT scans to correct signal loss in cardiac PET, in an arXiv paper posted on 19 Nov 2026.

The paper describes how LUCID corrects attenuation, the weakening of signals as they pass through tissue during positron emission tomography (PET). The model builds a correction map from the PET images without a separate computed tomography (CT) scan for each patient.

The researchers generated an initial correction map for each PET time frame. They used a diffusion model to refine each map in a single denoising step, then averaged the refined maps across frames.

The researchers evaluated LUCID on their own dataset of 102 patients. They reported that it achieved the highest quantitative accuracy among the methods they compared.

The paper also reported that when the diffusion model was pretrained on CT scans from the training patients, its refined maps were less accurate than the initial maps.

The abstract did not report validation results for scans from patients at other hospitals.