Adventures in modelling highly multiplexed imaging data
Kieran Campbell, PhD Investigator at the Lunenfeld-Tanenbaum Research Institute Assistant professor at the Departments of Molecular Genetics and Statistical Sciences University of Toronto
Highly multiplexed imaging data is a new technology that allows the simultaneous quantification of the expression of 30-50 proteins while retaining spatial information in the tissue of origin. This enables the quantification of multiple high-dimensional single-cell phenotypes and their corresponding spatial arrangements, which multiple studies have successfully linked to predict patient outcomes. However, issues of both experimental design and subsequent analysis can hamper biological interpretation. In this talk I will outline some of my lab’s recent work developing probabilistic clustering models that account for cellular segmentation errors, as well as outline the development of a machine learning powered web-app that enables antibody panel design for highly multiplexed imaging data.