Phantom-based evaluation of radiomics feature stability for low-dose CT lung cancer screening - Takeaways - MDSpire

Assessment of Radiomic Feature Consistency in Low-Dose CT Screening for Lung Cancer Using Phantom Models

  • By

  • Sunyi Zheng

  • Xiaomeng Yang

  • Hongren Wang

  • Zhipeng Gao

  • Pengchun Ye

  • Weiping Wang

  • Wenhua Li

  • Donghua Meng

  • Shuai Zhang

  • Wenjia Zhang

  • Houpu Liu

  • Shuyuan Huang

  • Chunlin Zhang

  • Jing Wang

  • Jihui Hao

  • Xiaonan Cui

  • July 21, 2026

Share

  • 1

    The study assessed the stability of radiomic features from lung nodules under low-dose CT screening conditions using phantom models.

  • 2

    Inter-scanner variability significantly impacted feature stability more than intra-scanner variability, with tube current being more influential than tube voltage.

  • 3

    Models using representative stable features outperformed those with unstable features in both malignancy assessment and nodule growth prediction.

  • 4

    The area under the receiver operating characteristic curve (AUC) for stable feature models was 0.995 for malignancy assessment and 0.799 for growth prediction.

  • 5

    Stable radiomic features were associated with improved performance consistency in independent lung cancer screening datasets.

Original Source(s)

Phantom-based evaluation of radiomics feature stability for low-dose CT lung cancer screening

  • by Sunyi Zheng, Xiaomeng Yang, Hongren Wang, Zhipeng Gao, Pengchun Ye, Weiping Wang, Wenhua Li, Donghua Meng, Shuai Zhang, Wenjia Zhang, Houpu Liu, Shuyuan Huang, Chunlin Zhang, Jing Wang, Jihui Hao, Xiaonan Cui

  • July 21, 2026

Related Content

Predictive modeling for survival-related outcomes in lung cancer patients with brain metastases: a mini-review

Consolidative versus salvage stereotactic ablative radiotherapy to the primary lung tumor in stage IV non–small cell lung cancer

Research progress of small molecule targeted drugs for non-small cell lung cancer