Histopathology Image Segmentation
Overview
The field develops methods to label tissue types and outline individual cells, nuclei, glands, or lesions in stained pathology images, from small image patches to gigapixel whole-slide images. Common approaches include U-Net-style convolutional networks, transformers and pathology foundation models, boundary-aware instance segmentation, and interactive or prompt-based models. Because pixel-level labels are costly, many methods use weak or semi-supervised learning, sparse annotations, pseudo-labels, or multimodal signals; performance is often tested on benchmarks such as PanNuke, MoNuSeg, GlaS, CRAG, and BCSS. A central challenge is maintaining accurate boundaries and reliable results across stains, hospitals, magnifications, and tissue types while reducing annotation and computation costs.
Latest papers
53 papers7 papers in the last four weeks, against 1 the four weeks before. 0.0% of all new papers.