. Cell Detection with Transformers – A Paradigm Shift from Segmentation to Detection in Digital Pathology. In 21st European Congress on Digital Pathology. 2025. 
(599.69 KB)
(599.69 KB) Abstract
Cell nuclei detection and classification are fundamental tasks in digital pathology, enabling biomarker quantification and tumor microenvironment analysis. Although the clinically relevant objective is to detect and classify individual cells, these tasks are typically addressed using segmentation-based methods. While effective, segmentation introduces significant computational overhead. To overcome these limitations, we propose CellNuc-DETR, a transformer-based detection model that directly detects and classifies cells, offering a more efficient and scalable alternative to segmentation-based approaches
