Predicting Invasiveness of Lung Adenocarcinoma from Chest CT with Few-shot Vision-Language Ternary Classification Model - Summary - MDSpire

Predicting Invasiveness of Lung Adenocarcinoma from Chest CT with Few-shot Vision-Language Ternary Classification Model

  • By

  • Nan Xu

  • Qianqian He

  • Lu Wang

  • Zhiwen Zhang

  • Qiuju Sheng

  • Shang Gao

  • Shimin Zhang

  • Bosinan Chen

  • Jianing Sun

  • Zhijian Zhang

  • Jie Zhang

  • Jing Qiu

  • Yunan Wang

  • Guanyu Liu

  • Enyu Li

  • Mingke Tian

  • Haotian Wang

  • Jiaping Yu

  • Yan Dong

  • Si Gao

  • Song Chen

  • Fan Yang

  • Zhihui Chang

  • Yue Dong

  • Lina Zhang

  • Jiangdian Song

  • December 20, 2025

  • 0 min

Share

Objective:

To investigate the potential of vision-language models in noninvasively predicting the invasiveness of pure ground-glass nodules (pGGNs) in lung adenocarcinoma using chest CT scans, focusing on the application of GPT-4o.

Key Findings:
  • The twenty-shot GPT-4o model outperformed other training strategies in diagnosing pGGN invasiveness (P < 0.01), with an AUC of 0.80 indicating good diagnostic capability.
  • GPT-4o achieved an accuracy of 75%, sensitivity of 74%, specificity of 86%, and an AUC of 0.80.
  • Radiologists showed improved diagnostic performance with GPT-4o assistance, indicating its potential to enhance clinical outcomes.
Interpretation:

The study demonstrates that GPT-4o can effectively assist radiologists in accurately classifying the invasiveness of pGGNs, potentially enhancing clinical decision-making and improving patient outcomes.

Limitations:
  • The study is retrospective and may have inherent biases, such as selection bias and data quality issues.
  • The performance of GPT-4o may vary with different datasets or imaging protocols.
Conclusion:

GPT-4o shows promise as a tool for improving the noninvasive assessment of pGGN invasiveness in lung adenocarcinoma, warranting further validation in clinical settings.

Original Source(s)

Predicting Invasiveness of Lung Adenocarcinoma from Chest CT with Few-shot Vision-Language Ternary Classification Model

  • by Nan Xu, Qianqian He, Lu Wang, Zhiwen Zhang, Qiuju Sheng, Shang Gao, Shimin Zhang, Bosinan Chen, Jianing Sun, Zhijian Zhang, Jie Zhang, Jing Qiu, Yunan Wang, Guanyu Liu, Enyu Li, Mingke Tian, Haotian Wang, Jiaping Yu, Yan Dong, Si Gao, Song Chen, Fan Yang, Zhihui Chang, Yue Dong, Lina Zhang, Jiangdian Song

  • December 20, 2025

Related Content

Real-World Outcomes of Immune Checkpoint Inhibitors in Lung Cancer: A Study from a Middle-Income Nation

Evidential deep learning-based ALK-expression screening using H&E-stained histopathological images

Enhancing Access to Lung Cancer Screening with Emphasis on Health Disparities: Insights from Professionals in the UK NHS Lung Cancer Screening Initiative