Multi-parametric diffusion-weighted magnetic resonance imaging and microstructural tissue properties in breast cancer

Abstract

Breast cancer is the most common cancer in women worldwide. Although breast cancer survival has improved in recent years, it remains a leading cause of cancer death in the female population worldwide. Breast cancer prognosis is strongly associated with the staging of the cancer at the time of diagnosis. Today, newly diagnosed early-stage breast cancer has a high 5-year survival rate in developed countries. For the most advanced cancers however, the survival rates are much lower. Due to the heterogeneity and complexity of breast cancer biology, patients with similar clinical diagnosis can have very different prognosis and response to treatment. There is an urgent need for more efficient and safer breast cancer therapeutic regimens based on individual tumor properties. In parallel, early detection and defining new strategies for patient stratification for prediction and monitoring of treatment response are warranted.

Radiographic x-ray mammography is the gold standard for breast cancer screening. The standard magnetic resonance imaging (MRI) techniques dynamic contrast-enhanced (DCE) and anatomic T2 weighted imaging have been incorporated into the BI-RADS with high sensitivity and variable specificity. Diffusion MRI has been suggested as a complementary technique to DCE to improve diagnostic accuracy, in particular with respect to specificity, and could potentially reduce unnecessary biopsies. Further, diffusion weighted imaging (DWI) may help to stratify breast cancers by subtypes or prognostic factors which are important for treatment planning.

This thesis aims to further explore the benefits of diffusion-weighted MRI in breast cancer applications. The main goal was to gain insight into the potential of advanced diffusion models and how they relate to tissue microstructure in breast. The specific aims include: to explore the underlying tissue properties and structural changes in breast cancers that give rise to the contrast observed in diffusion-weighted MRI, to evaluate advanced non-Gaussian models for the diffusion-weighted signal in the high b-value range 200 - 3000 s/mm2 for breast cancer classification, to evaluate a machine-learning algorithm for breast cancer classification and subtyping using diffusion histogram properties, and to investigate the effects of echo time dependency in DWI measurements.