Deep learning–based CT slice synthesis improves radiomic feature reproducibility and discriminative performance in lung nodule assessment
Hujun Yang, Zhengping Zhang, Lei Tian, Pei Dang, Long Wan, Zhiyuan Zhou, Jingjie Li, Lei Cao, Xiaoyan Yang, Juan Chen · 2026
Read Deep learning–based CT slice synthesis improves radiomic feature reproducibility and discriminative performance in lung nodule assessment by Hujun Yang, Zhengping Zhang, Lei Tian, Pei Dang, Long Wan, Zhiyuan Zhou, Jingjie Li, Lei Cao, Xiaoyan Yang, Juan Chen (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/PMC13282418.txt.
Description
Hujun Yang and co-authors report a 2026 study on using deep learning to fix a slice thickness problem in lung scan analysis. Radiomics extracts many numeric features from CT scans to judge whether lung nodules are benign or cancerous. Thick slices make those features unstable across hospitals. The team studied five hundred six patients from two cohorts, with scans rebuilt at one and five millimetres.
An algorithm converted thick-slice scans into synthetic thin-slice scans. Features from synthetic scans agreed far better across thicknesses than features from real thick scans. Only about one percent of thick-scan features met the reproducibility bar, against about twenty-seven percent after synthesis. Models built on synthetic scans also separated benign from cancerous nodules better than thick-scan models, matching real thin-scan models. The authors present synthesis as a practical standardisation step.
Catalog record
| CREATOR | Hujun Yang, Zhengping Zhang, Lei Tian, Pei Dang, Long Wan, Zhiyuan Zhou, Jingjie Li, Lei Cao, Xiaoyan Yang, Juan Chen |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 44988 |
| LICENSE | https://creativecommons.org/licenses/by/4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |