Related Experiment Video
Updated: Oct 6, 2025
10:17
Guidelines and Experience Using Imaging Biomarker Explorer IBEX for Radiomics
Published on: January 8, 2018
13.3K
[The stability and repeatability of radiomics features based on lung diffusion-weighted imaging]
1Department of Radiology, the First Affiliated Hospital of Guangzhou Medical University, Guangzhou 510120, China.
Zhonghua Yi Xue Za Zhi
|January 18, 2022
Summary
Radiomics features from lung diffusion-weighted imaging (DWI) show good robustness and reproducibility. These imaging biomarkers are reliable for clinical use, with specific sequences and features demonstrating higher stability across scans and observers.
Area of Science:
- Radiology
- Medical Imaging
- Quantitative Imaging
Background:
- Radiomics analysis of lung diffusion-weighted imaging (DWI) is emerging for pulmonary nodule characterization.
- Evaluating the feasibility, robustness, and reproducibility of radiomics features is crucial for clinical translation.
Purpose of the Study:
- To assess the feasibility, robustness, and reproducibility of radiomics features extracted from lung DWI.
- To compare the performance of different DWI sequences (EPI and TSE) and apparent diffusion coefficient (ADC) maps.
Main Methods:
- Prospective collection of DWI scans from 30 patients with pulmonary nodules/masses.
- Acquisition of both Echo Planar Imaging (EPI) and Fast Spin-Echo (TSE) DWI sequences, repeated for robustness assessment.
- Manual segmentation by two radiologists and extraction of 396 radiomics features, analyzed for robustness (CCC, DR) and reproducibility (ICC).
Main Results:
- A significant proportion of radiomics features demonstrated robustness and reproducibility across different scans and observers.
- DWI sequences yielded a higher number of robust features compared to ADC maps.
- 54 out of 396 features (13.6%) showed excellent robustness and reproducibility across all tested sequences.
Conclusions:
- Radiomics features derived from lung DWI are robust and reproducible.
- The choice of DWI sequence and feature type influences stability.
- Certain radiomics features exhibit reliable performance across different scans, observers, and sequences, supporting their potential clinical utility.
