Multi-marker quantitative radiomics for mass characterization in dedicated breast CT imaging

Area of Science:

• Radiology
• Medical Imaging
• Computational Pathology

Background:

• Dedicated breast computed tomography (bCT) is an emerging imaging modality for breast cancer detection.
• Accurate classification of breast masses as benign or malignant is crucial for patient management.
• Radiomics offers a quantitative approach to extract imaging features for diagnostic purposes.

Purpose of the Study:

• To develop and evaluate a multi-marker radiomic algorithm for classifying breast masses on bCT images.
• To assess the diagnostic performance of the developed radiomic model.

Main Methods:

• Development of over 1000 radiomic descriptors quantifying mass heterogeneity, morphology, and margin characteristics.
• Feature selection using stability analysis, statistical significance, and interaction evaluation.
• Training a linear discriminant analysis (LDA) radiomic model on 202 bCT masses and testing on an independent set of 82 cases.

Main Results:

• Individual radiomic descriptors showed diagnostic performance (AUC > 0.65) on the training set.
• The final LDA radiomic model achieved an AUC of 0.90 (95% C.I. 0.80-0.96) on the independent test set.

Conclusions:

• The multi-marker radiomic approach demonstrates high diagnostic accuracy for breast mass classification in bCT.
• Quantitative radiomics applied to bCT has the potential to enhance the breast cancer diagnostic pipeline.
• Further validation with larger datasets is warranted.