Fourier Analysis of Bilateral Breast Asymmetry for Short-term Breast Cancer Risk Prediction

Area of Science:

• Medical Imaging
• Radiology
• Breast Cancer Research

Background:

• Mammogram measures are established predictors of breast cancer risk.
• Bilateral breast asymmetry analysis offers potential for short-term risk prediction.

Purpose of the Study:

• To investigate bilateral breast asymmetry from mammograms as a short-term breast cancer risk predictor.
• To compare the efficacy of raw full-field digital mammography (FFDM) versus processed FFDM and digital breast tomosynthesis (DBT) images.

Main Methods:

• A matched case-control study analyzing mammograms acquired near diagnosis.
• Fourier domain (FD) analysis to quantify breast asymmetry from contralateral images.
• Conditional logistic regression modeling to calculate odds ratios (ORs) and area under the receiver operating curves (Azs).

Main Results:

• Raw FFDM images showed significant breast cancer risk prediction (OR=1.90, Az=0.72).
• Processed FFDM and DBT images demonstrated attenuated associations (ORs 1.31 and 1.48, Azs 0.63 and 0.65, respectively).
• DBT images, despite lower resolution, yielded stronger associations than clinical FFDM images.

Conclusions:

• Clinical processing of FFDM and DBT images leads to information loss for breast cancer risk prediction.
• Raw FFDM images are superior to processed images and DBT for capturing asymmetry relevant to risk.
• Breast asymmetry analysis using FD holds promise for mammographic risk assessment.