Predicting mammographic density with linear ultrasound transducers

Area of Science:

• Medical Imaging
• Radiology
• Oncology

Background:

• High mammographic density (MD) is a significant risk factor for breast cancer (BC).
• Factors influencing MD include age, BMI, pregnancies, and lactation.
• Non-radiation imaging alternatives to mammography, like ultrasound, are crucial for MD assessment.

Purpose of the Study:

• To evaluate the predictive power of ultrasound imaging for percentage mammographic density (PMD).
• To explore alternative, non-radiation methods for assessing breast density.

Main Methods:

• 168 patients underwent mammography and high-frequency ultrasound (12 MHz).
• Ultrasound image gray level bins were used to characterize MD.
• Regression models, including random forest, predicted PMD from gray level bins.

Main Results:

• A limited correlation was observed between gray level bins and PMD (Spearman's ρ: -0.18 to 0.32).
• The random forest model showed the highest accuracy (cross-validated R²: 0.255).
• Ultrasound imaging with the VOLUSON® 730 Expert demonstrated limited predictive power for PMD.

Conclusions:

• No reliable prediction of PMD using the evaluated ultrasound imaging method was achieved.
• The predictive power of ultrasound for MD appears device-dependent.
• Further research into non-radiation breast imaging methods is essential.