Project 1: MRI attributes for predictive models in oncology

Background

Conventional Magnetic Resonance Imaging (MRI) supports tumor diagnosis, treatment selection, and treatment monitoring by evaluating lesion location, size, and structure. Advanced methods of MRI additionally provide insides into tumor angiogenesis, cellularity, and hypoxia, which potentially prognosticate disease course and treatment response. 

In this project, we investigate and select information on lesion morphology, structure, and aggressiveness from MRI scans. Our goal is the development of optimal MRI scanning procedures that allow fast and accurate supports to oncological treatment.

Working hypothesis

  1. MRI offers a multiplicity of patient- and lesion-specific imaging attributes, which support automated tissue classification and identification of imaging prognostic markers;
  2. The analysis of the reliability, specificity, and sensitivity of those MRI features underlies the development of algorithms for accurate, fast, and operator-independent classification of tumor types and prediction of disease course.

Specific aims for this research project

  1. To implement multi-parametrical and clinically feasible MRI protocols for evaluation of tumor morphology and pathophysiology of malignant lesions in well-defined oncological sub-cohorts;
  2. To provide relative weights of MRI-derived features for tissue classification and prognosis of disease course;
  3. To improve the prognostic value of the MR-based predictions by integrating multi-modality data.
Image Project 1