Standardized CT Preprocessing Pipeline Harmonizes TCIA Data for Radiomics and AI

A reproducible CT preprocessing framework for The Cancer Imaging Archive harmonizes orientation, spacing, and intensities into analysis-ready volumes, reducing non-biological variability that can degrade radiomic feature stability and AI model performance.

• Technical development and demonstration using TCGA-KIRC CT data from The Cancer Imaging Archive (TCIA); no patient-level outcomes reported.
• The modular pipeline includes DICOM-to-NIfTI conversion, orientation harmonization, voxel resampling, intensity normalization, and segmentation integration.
• Not yet externally validated; improvement in downstream radiomics or AI task performance remains unquantified.

RadPigeon summaries are original and for information only. They are not clinical advice.