SUMI: Distilling Photon-Counting CT Quality into Routine Chest CT via Degradation Modeling

Summary

Photon-counting CT (PCCT) offers higher spatial resolution and lower noise than conventional energy-integrating CT (EICT), but its limited clinical availability restricts large-scale research. This paper introduces SUMI, a simulate-degrade-then-enhance approach that learns to reverse realistic acquisition artifacts in low-quality EICT scans. The authors model the degradation from PCCT to routine chest CT and train an enhancement network that restores PCCT-like quality from standard CT images. On external validation data, SUMI outperforms state-of-the-art image enhancement methods by 15% in SSIM and 20% in PSNR. Beyond image-quality metrics, the enhanced scans improve radiologists' subjective clinical utility ratings and boost downstream lesion detection performance. The work, posted to arXiv (2504.06848) in the computer vision category by Junqi Liu, Xinze Zhou, and Wenxuan Li, suggests a practical route to access PCCT-level diagnostic quality using widely deployed routine CT scanners.