Breast cancer is the most prevalent cancer in women. Detection of early invasive breast cancers plays an important role in cancer treatment and reduction of patient mortality rate. Studies have demonstrated that breast magnetic resonance imaging (MRI) was the most sensitive imaging modality to detect early breast cancer. However, since the cancer detection yield of applying breast MRI to the general screening population to detect mammography-occult early cancers is quite low (~1%), current ACS guideline limits breast MRI screening examinations to a small group of women with elevated cancer risk (e.g., lifetime risk > 20%) determined by several epidemiology study based risk prediction models. As a result, although current breast MRI screening has 2 to 3% cancer detection yield, which is still too low to be cost-effective, the majority of women harboring mammography-occult breast cancers is excluded from the breast MRI screening. In order to significantly increase cancer detection yield of breast MRI screening, we propose to develop and test a new strategy and risk models that emphasize on predicting the risk of a woman having mammography-occult cancer in a short-term after a negative mammography and/or MRI screening examination. Our new models will be developed based on the quantitative analysis of image features related to the bilateral asymmetry of mammographic tissue density and/or MRI background parenchymal enhancement between the left and right breasts of the same woman, which is highly correlated to the biological process of cancer development. Our goal is to help significantly increase cancer detection yield of the breast MRI screening by combining two new risk assessment approaches. The first one is a rule-in approach that uses a new risk model to identify women with higher risk of harboring mammography-occult early cancer from the general screening populations. The second one is a rule-out approach that use another new risk model to identify women who currently are recommended for breast MRI screening examinations due to the elevated lifetime cancer risk but actually do not have imminent risk of having breast cancer, so that many of these women can be screened in a longer interval to reduce the repeated negative screening examinations until their short-term risk scores are increased significantly in the future assessments. To test the feasibility of this new strategy and risk models for breast MRI screening, we will develop and optimize new quantitative image feature detection schemes to identify new image biomarkers and build new short-term cancer risk prediction models. We will conduct both retrospective and prospective studies to test the performance and reliability of our new risk models. The goal or milestone of this project is to increase cancer detection yield of breast MRI screening from ~2% when applying only to a limited group of women to greater than 10% when applying to the general mammography screening population. If successful, this project will have a high clinical impact in helping improve efficacy of breast cancer screening by applying breast MRI screening to detect more mammography-occult breast cancers at an early stage.
This project aims to develop and test a new strategy and two short-term cancer risk prediction models to increase cancer detection yield of breast MRI screening. Based on the quantitative analysis of image features computed from bilateral mammograms and/or MRI images of the left and right breasts, the new models can be applied to general mammography screening population to identify the women with high risk of harboring mammography-occult early cancer and also help determine the optimal MRI screening interval of women with elevated breast cancer risk determined by existing risk models.
Showing the most recent 10 out of 17 publications
