Research

Breast tumor heterogeneity is a recognized factor correlated to type and aggressiveness. For example, cancers are known to express a higher degree of intra- and peri-tumoral heterogeneity compared to benign tumors, due to abnormal cellular proliferation and vascular remodeling. Among cancers, it is hypothesized that a higher level of intra-tumor heterogeneity, due to multiple cellular lines composing the tumoral environment, may yield higher resistance to treatment.

To depict tumor heterogeneity, functional imaging could play a key role, since it is non-invasive, cost-effective, and allows to analyze the tumor in its entirety. In functional imaging, images are acquired at different temporal moments after the intravenous administration of contrast-enhancing material, allowing for the observation of how the tumor enhances differently in space (i.e. throughout and around the tumoral region) and time (i.e. how different areas of the tumor activate differently over post-contrast phases).

To move one step further and provide a quantitative characterization of tumor heterogeneity (and not only depict it through visual assessment), radiomics could play a key role. Radiomics is a relatively new image analysis paradigm aimed at the extraction of quantitative, clinically relevant information from medical images. This extracted information is used to develop Artificial Intelligence models for clinical decision support, for example to predict tumor type, grade, and response to treatment.

In this project, we are developing advanced radiomics algorithms to quantify tumor heterogeneity from dynamic contrast enhanced magnetic resonance imaging (DCE-MRI), to extract relevant imaging biomarkers that can be used to improve diagnosis, prognosis, and treatment response prediction. We are especially focusing on quantifying heterogeneity in four dimensions (4D, 3D over time), to leverage the spatial and temporal image information made available by DCE-MRI.

Researchers:

Ioannis Sechopoulos

Ritse Mann