Repeatability of Multiparametric Prostate MRI Radiomics Features.

Area of Science:

• Radiology
• Medical Imaging
• Oncology

Background:

• Radiomics offers quantitative image-based biomarkers for disease detection and characterization.
• Repeatability of radiomics features is essential for their clinical utility as biomarkers.
• Prostate cancer diagnosis benefits from accurate and reproducible imaging biomarkers.

Purpose of the Study:

• To assess the repeatability of radiomics features for small prostate tumors using test-retest multiparametric MRI (mpMRI).
• To investigate the impact of various preprocessing and extraction configurations on radiomics feature repeatability.
• To provide recommendations for improving the reliability and reporting of radiomics studies.

Main Methods:

• Test-retest multiparametric MRI (mpMRI) scans were acquired for small prostate tumors.
• Radiomics features were extracted under diverse preprocessing configurations, including normalization, pre-filtering, and discretization (bin width).
• Repeatability was quantified using the Intraclass Correlation Coefficient (ICC).

Main Results:

• Many radiomics features and preprocessing combinations demonstrated high repeatability (ICC > 0.85).
• Overall feature repeatability was found to be highly sensitive to specific processing parameter choices.
• Image normalization and pre-filtering did not consistently enhance repeatability across different configurations.

Conclusions:

• Radiomics feature repeatability in prostate cancer is significantly influenced by processing parameters.
• Caution is advised when interpreting radiomics features; detailed reporting of processing configurations is essential.
• Open-source implementations are recommended to promote transparency and reproducibility in radiomics research.