Breast Cancer Risk in Women with Breast Bilateral Asymmetry: Machine Learning Based Risk Analysis and Mitigation through Developing a Framework for Customized Bra Design
Abstract
Breast cancer is the most prevalent form of cancer globally, accounting for 12.5% of all new cases annually. Research has found a significant correlation between breast bilateral asymmetry and an increased risk of cancer, with women diagnosed with breast cancer having higher levels of bilateral asymmetrical breast volume. Unfortunately, 87% of women with breast asymmetry lack adequate tools for assessing their cancer risk. Early screening using bilateral asymmetry to predict a woman's long-term risk of breast cancer can help physicians make informed decisions about whether to recommend sequential imaging and the frequency of screening. Another important factor in understanding the cause of breast cancer is the association between long-term abnormal mechanical stress distribution in breast tissue and the increased risk of developing breast lesions. Chronic stress promotes cancer development through various molecular mechanisms. However, existing off-the-shelf symmetric bras do not adequately address breast asymmetry, as they may not provide sufficient support for smaller breasts while inducing high stress levels to larger breasts. Therefore, it is essential to explore the relationship between concentrated stress from ill-fitted bras and its potential contribution to breast cancer development. A more personalized and tailored bra fitting technique could significantly reduce the risk of breast cancer associated with mechanical stress. In this study, we developed an unsupervised machine learning algorithm to classify breast bilateral asymmetry using bilateral magnetic resonance imaging. A clear link between breast asymmetry and breast cancer risk has been established, providing a predictive tool for proactive breast health assessment. We then developed two complementary computational inversion techniques to determine the individual-specific hyperelastic parameters of breast tissue, along with the breast's undeformed shape, using MRI images. This synergistic algorithm addresses issues with preloading-induced errors, thereby providing a more precise foundation for designing customized bras. The development of customized bras for cancer-prone women with significant breast asymmetry is facilitated by the optimization of breast tissue stress distribution. This is achieved through the accurate capture of breast shape and tissue properties. By integrating these details with various textile options for bra modeling, our study supports the natural state of the breast and reduces potentially harmful stress concentrations. Our research contributes significantly to the understanding of breast cancer risk factors and offers potential for innovative approaches in preventive breast healthcare. This study is a crucial step forward in the field and demonstrates the potential for improved outcomes in breast health.