CVPR Daily - Wednesday

Histopathology whole slide image analysis plays a crucial role in diagnosing cancer, determining its stage, and identifying its specific type. However, the challenge of processing large gigapixel-sized images using traditional methods has hindered progress in the field, as these images cannot be easily accommodated in memory. In this paper, Howard and Fernando propose an innovative approach to cancer detection using advanced graph-based techniques for analyzing Howard Chan Tsai Hor (left) is a second-year PhD student, and Fernando Julio Cendra (right) is a master’s student at the University of Hong Kong. Their paper proposes a novel graph data structure to model histopathology images. They speak to us ahead of their poster this afternoon. Histopathology Whole Slide Image Analysis with Heterogeneous Graph Representation Learning 14 DAILY CVPR Wednesday Poster Presentation Whole slide image illustration and our constructed heterogeneous graph