Clinical Report: Assessment of Radiomic Feature Consistency in Low-Dose CT

Overview

This study evaluates the stability of radiomic features from lung nodules under low-dose CT conditions. Findings indicate that inter-scanner variability significantly affects feature stability.

Background

Lung cancer is the leading cause of cancer-related mortality globally, with low-dose computed tomography (LDCT) screening showing a significant reduction in mortality rates. Despite the benefits of LDCT, challenges in accurately characterizing pulmonary nodules persist, necessitating improved quantitative imaging methods. Radiomics, which quantifies imaging data into high-dimensional features, has emerged as a promising approach for enhancing lung cancer screening.

Data Highlights

Key Findings

• Inter-scanner variability has a greater impact on feature stability than intra-scanner variability.
• Tube current influences feature stability more than tube voltage.
• Models using representative stable features outperformed those using unstable features in both malignancy assessment and nodule growth prediction.
• Stable feature models exhibited smaller performance differences between training and test sets compared to unstable feature models.
• High stability of radiomic features correlates with improved performance consistency in independent datasets.

Clinical Implications

The findings indicate that selecting stable radiomic features may enhance the reliability of lung cancer screening models.

Conclusion

Stable radiomic features are associated with improved model performance in lung cancer screening.

Related Resources & Content

1. European Radiology, 2023 -- Assessment of Radiomic Feature Consistency in Low-Dose CT Screening for Lung Cancer Using Phantom Models
2. European Radiology (Springer) — Comparing the detectability of pulmonary nodules on two ultra-high resolution CT scanners: a preliminary phantom study
3. European Radiology — Evaluating the Stability of Radiomics in Photon-Counting Detector CT: Effects of Acquisition and Reconstruction Variables
4. European Radiology — Assessment of Test-Retest Stability in Radiomics Features Using Phantom-Based Photon-Counting Detector CT
5. Lung Cancer: Screening | United States Preventive Services Taskforce
6. Reduced Lung-Cancer Mortality with Low-Dose Computed Tomographic Screening
7. The Image Biomarker Standardization Initiative: Standardized Quantitative Radiomics for High-Throughput Image-based Phenotyping - PMC