The detection of tumor tissue in histological sections of breast cancer biopsies is important to achieve a correct quantification of aggressiveness markers, such as Ki67 (cell proliferation marker), which is involved in the diagnosis and determination of the treatment after surgery. This detection is usually done by manual annotation. The goal is to automate this delineation by deep learning. This delineation must be robust to changes in immunohistochemical labeling, so that it can be applied to other markers of interest. A secondary objective will be to detect "hot spots" (areas of high density of Ki67-positive cell nuclei) of proliferation within the tumor area.
This study will be made in collaboration with Prof. X. Catteau (CurePath, Centre universitaire inter-régional d'expertise en anatomie pathologique hospitalière, Jumet).

Prerequisite:
Programming (Python), machine learning and image processing.

Contacts: Egor Zindy (egor.zindy_at_ulb.ac.be), Christine Decaestecker (cdecaes_at_ulb.ac.be)