Scalable Visual Analytics for Digital Cancer Pathology, Robert Krueger
Sprache des Titels:
Englisch
Original Kurzfassung:
With new tumor imaging technologies, cancer biology has entered a digital era. Artificial intelligence has enabled the processing and analysis of imaging data at unprecedented scale. While processing pipelines are rapidly evolving, pre-clinical research performed with the data is highly experimental and exploratory in nature, making integration of biomedical experts essential to steer workflows and interpret results.
In my talk, I will introduce a scalable rendering framework enabling users to load, display, and interactively navigate terabyte-sized multiplexed images of cancer tissue. I will then present visual analytics interfaces that build on this framework and support cell biologists and pathologists in their workflows. By leveraging both unsupervised and supervised learning in an interactive setting, cells in the tissue can be iteratively classified into tumor, immune, and stromal cell type hierarchies. Subsequently, spatial neighborhoods of cells are quantified in order to query and cluster reoccurring, biologically-meaningful cellular interactions both in and across specimens. Once relevant biological patterns are identified, a novel focus and context lensing interface enables pathologists to further assess and annotate these regions of interest in an intuitive fashion. I will conclude with an outlook into my future research agenda, addressing the transition to volumetric and time-varying datasets, detailed analysis of cell-cell interaction profiles in high-resolution 3D data, and the joint exploration of multimodal images with increasing amounts of spatially-referenced sequencing data.
Forschungs-