Fusion Annotator: A Platform for Accelerating Consensus-Driven Ground Truth Generation with AI Assistance

Artificial intelligence now plays a central role in computational pathology, enabling large-scale analysis of whole slide images for segmentation, classification, and quantitative feature extraction. Despite these advances, the development of robust and generalizable models remains constrained by the limited availability of high-quality, expertly annotated datasets. This limitation is particularly acute in renal pathology, where ground-truth generation is challenged by labor-intensive workflows, limited subspecialty expertise, inter-observer variability, heterogeneous imaging protocols, and restricted data accessibility. To address these challenges, we present Fusion Annotator, a cloud-first, pathologist-centered digital annotation platform designed to accelerate and standardize expert-driven dataset creation through consensus-oriented workflows. Unlike conventional desktop-based annotation tools that require local installation and hinder collaborative scalability, Fusion Annotator operates entirely within a web browser, enabling secure multi-institutional access, real-time updates, and centralized data management. The platform integrates with the Computational Renal Pathology Suite, leveraging open-source machine learning models to pre-identify regions of interest and functional tissue units, thereby reducing manual annotation burden. Fusion Annotator supports customizable descriptor schemas tailored to specific disease cohorts and study objectives, enabling structured and flexible annotation protocols. Additional features include reference image guidance to support annotator training, multi-expert annotation to facilitate consensus building, and iterative refinement informed by continuous feedback from practicing renal pathologists. The platform is currently being used to generate expert-annotated datasets for focal segmental glomerulosclerosis, diabetic nephropathy, and renal transplant cohorts. Fusion Annotator provides an extensible pathway for producing high-quality renal pathology datasets to support scalable AI development in computational pathology.