Cascaded deep learning for automated lung cancer tumour burden quantification on [18F]FDG PET/CT
Shaonan Zhong, Jie Lv, Zhaohong Pan, Youcai Li, Yimin Fu, Bingsheng Huang, Xinlu Wang · 2026
Read Cascaded deep learning for automated lung cancer tumour burden quantification on [18F]FDG PET/CT by Shaonan Zhong, Jie Lv, Zhaohong Pan, Youcai Li, Yimin Fu, Bingsheng Huang, Xinlu Wang (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/PMC13457853.txt.
Description
Shaonan Zhong and colleagues build a two-stage deep learning system that measures lung cancer burden on FDG PET/CT scans. Manual outlining of hot spots is slow, and single-stage segmentation flags false lesions. Their 2026 retrospective study gathers 593 scans from one center across two scanners: 496 for cross-validated training and internal testing, 97 held out as an external test.
The cascade first segments candidate lesions, then runs a classification stage that weeds out false positives before computing burden. Accuracy held at a Dice score of 0.82 on both test sets, with precision significantly better than segmentation alone. Automated volumes tracked the human references closely, correlating at 0.984 for metabolic tumor volume and 0.998 for total lesion glycolysis. The authors pitch the system as a reliable way to quantify tumor burden for diagnosis, prognosis, and treatment planning, cutting radiologist workload while keeping numbers trustworthy.
Catalog record
| CREATOR | Shaonan Zhong, Jie Lv, Zhaohong Pan, Youcai Li, Yimin Fu, Bingsheng Huang, Xinlu Wang |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 28490 |
| LICENSE | https://creativecommons.org/licenses/by/4.0 |
| RIGHTS | CC BY |
| SOURCE | Europe PMC |