Cross-Software Radiomic Feature Robustness Assessed by Hierarchical Clustering and Composite Index Analysis: A...

Area of Science:

• Medical Imaging
• Radiomics
• Computational Pathology

Background:

• Radiomic feature robustness is essential for reproducible imaging biomarkers.
• Software variability can compromise feature consistency and predictive model reliability.
• Clinical translation of radiomic biomarkers requires stable, cross-platform features.

Purpose of the Study:

• To develop and validate a hierarchical clustering workflow for assessing radiomic feature robustness.
• To identify stable and reproducible radiomic features across different software platforms.
• To establish a methodological foundation for cross-platform harmonization of radiomic biomarkers.

Main Methods:

• Analysis of a multi-cancer CT dataset (colorectal cancer, liver metastases, hepatocellular carcinoma).
• Extraction of radiomic features using two IBSI-compliant software platforms.
• Assessment of intra-software reliability (ICC) and cross-software stability (hierarchical clustering, ARI).
• Quantification of inter-platform robustness using a Composite Index (CI).

Main Results:

• Over 95% of features showed good-to-excellent intra-software reliability.
• Hierarchical clustering achieved perfect concordance (ARI=1.0) across platforms.
• Wavelet-derived descriptors and first-order statistics (e.g., cluster shade, mean intensity) were most robust.

Conclusions:

• The proposed framework effectively identifies stable, transferable radiomic features across IBSI-compliant platforms.
• This methodology supports cross-platform harmonization for radiomic biomarkers.
• Findings enhance the reproducibility of radiomic biomarkers in oncologic imaging.