Dr. Sarder presented a talk titled “Computational Annotations of Cells in Histology using AI” at the School of Biomedical Engineering, University of British Columbia, Vancouver, Canada.
This talk started with giving an overview of challenges of ground-truth annotation in the detection and estimation of cells and associated states from brightfield histology digital images of thin tissue sections. Next we discussed how recent advancement in high resolution microscopy and spatial omics technologies has opened up opportunities to exploit the resulting imaging omics data as ground-truth labels for cell and associated state identification from brightfield histology images using artificial intelligence (AI) tools. With respect to advantages, brightfield microscopy histology is inexpensive and has served, for decades, as a “gold-standard” in pathology. We then discussed mapping of spatial omics data on brightfield histology using AI tools, using examples from two existing spatial omics technologies, namely co-detection by indexing (CODEX) and spatial transcriptomics. Utility of resulting fused data linking spatial omics and histology was discussed. Approaches to disburse the resulting pipelines for end-users via cloud based web-portal or desktop software was discussed. We concluded via discussing educational plans to train next generation scientists on challenges involved in cell annotation in pathology. Examples from renal pathology were used to guide the talk throughout the presentation.