Development of a phantom to test fully automated breast density software - A work in progress

Area of Science:

• Biomedical Imaging
• Medical Physics
• Radiology

Background:

• Mammographic density (MD) is a significant independent risk factor for breast cancer.
• Automated software can estimate MD, potentially enabling stratified screening.
• The precise relationship between breast thickness reduction and MD estimation requires further investigation.

Purpose of the Study:

• To develop a novel deformable breast phantom for evaluating automated mammographic density software.
• To assess the impact of breast thickness reduction on mammographic density measurements using the phantom.
• To compare phantom properties with those of real human breasts.

Main Methods:

• Poly vinyl alcohol (PVAL) phantoms with varying configurations were created.
• Mammographic image data was analyzed using Volpara software for volumetric breast density (VBD%).
• Hounsfield units (HU) were measured via CT scans, and physical density (g/cm³) was calculated.

Main Results:

• The Volpara software successfully identified all created deformable phantoms as female breasts.
• Phantom thickness reduction led to alterations in estimated phantom density.
• The phantoms could not withstand the same compression levels and thickness reduction as female breasts during mammography.

Conclusions:

• The developed deformable phantoms provide promising, valid data for automated breast density measurement.
• This study represents the first instance of creating deformable phantoms recognized as breasts by Volpara software.
• Further research is needed to create phantoms from PVAL and other materials that better mimic female breast structure, density, and compressibility.