Assessment of proliferation is an important step in the prognostic evaluation of breast cancer. Alongside other prognostic markers, measuring proliferation can give insight on the likely outcome of a patient and whether or not the patient should receive chemotherapy. Currently three prognostic markers, as described below, can measure proliferation. The aim of this work package is to use artificial intelligence to develop tools to measure, standardize and improve reproducibility and accuracy of this prognostic assessment.
Aim 1: Develop a tool to quantify mitotic activity
Counting the number of actively dividing cells, mitoses (as labelled in the image below), in a fixed region.
Aim
2: Develop tools to quantify Ki-67 and Phosphohistone protein 3 (PPH3/PHH3)
PPH3: Counting the number of positively stained tumour nuclei (brown) in a fixed region
Ki-67: Calculating the percentage of positively stained tumour nuclei (brown) in a hotspot
The image aboves shows the workflow of a machine-learning based tool for assessment of Ki-67 or PPH3. Top left, detection of breast tissue. Top middle, detection of the most proliferative region using a heatmap. Top right, identification and labelling of positive (blue) and negative (green) nuclei, and other cells (pink). The percentage Ki67 or number of PPH3 positive cells can now be measured within the specified area.