3D radiomics profiling of thyroid tumors using micro-CT

Area of Science:

• Radiology
• Pathology
• Oncology

Background:

• Tumor heterogeneity impacts cancer progression and treatment, often missed by 2D histology.
• 3D assessment of tumor microarchitecture is crucial for understanding spatial complexity.
• Micro-computed tomography (Micro-CT) offers high-resolution 3D virtual histology for soft tissues.

Purpose of the Study:

• To evaluate the utility of Micro-CT radiomics for quantitative 3D characterization of thyroid tumors.
• To assess the ability of Micro-CT radiomics to classify thyroid tissues and identify specific mutations.
• To explore potential imaging biomarkers for TERT promoter mutations.

Main Methods:

• Analysis of radiomics signatures from Micro-CT imaging of 418 thyroid tumor tissue microarrays.
• Classification of neoplastic vs. non-neoplastic tissues, papillary vs. follicular thyroid neoplasms, and BRAF V600E mutation status.
• Utilized Shapley additive explanations to identify key visual features driving classification and exploratory analysis for TERT mutations.

Main Results:

• Achieved robust classification of neoplastic vs. non-neoplastic thyroid tissues.
• Successfully differentiated between papillary thyroid carcinoma and follicular thyroid neoplasm.
• Demonstrated accurate classification of BRAF V600E mutation status and identified potential radiomics patterns for TERT promoter mutations.

Conclusions:

• Micro-CT radiomics shows promise as a complementary tool for thyroid cancer diagnostic classification.
• This technique enables quantitative 3D tissue characterization beyond visual inspection.
• Further validation is needed, but Micro-CT radiomics offers a promising platform for enhanced cancer diagnostics.