An automated toolbox for microcalcification cluster modeling for mammographic imaging

Area of Science:

• Medical Imaging
• Radiology
• Computational Modeling

Background:

• Mammography is crucial for breast cancer detection, relying on accurate interpretation of microcalcification clusters.
• Subtle variations and the sheer number of calcifications present interpretation challenges, even with advanced imaging techniques.

Purpose of the Study:

• To introduce an automated, flexible toolbox for generating customized microcalcification cluster models.
• To address limitations in existing simulation methods that often fail to capture lesion diversity or case-specific relevance.

Main Methods:

• The toolbox enables control over lesion size, calcification shape, and cluster density for diverse model generation.
• Two methods are presented: 3D cluster modeling using geometric shapes and 2D cluster modeling with radiomics analysis for tissue integration.
• The toolbox is Python-based, accessible via Jupyter Notebook, and includes validation by radiologists.

Main Results:

• Demonstrated flexibility in generating diverse microcalcification models using multiple simulation methods.
• Validated capacity to simulate realistic clusters with clinical properties, suitable for various mammography tasks and systems.
• Confirmed straightforward and fast model generation through automation.

Conclusions:

• The simulation toolbox provides a flexible platform for technical and clinical mammography research.
• Offers advanced 3D modeling for calcification characteristics and novel 2D modeling for texture integration.
• Facilitates the creation of tailored microcalcification models for improved imaging analysis and AI training.