CellDX

CellDX is a digital pathology platform developed by HistAI, designed for pathologists, researchers, and pharma and biotech organizations. It combines access to a large curated whole slide image (WSI) dataset with cloud-based slide viewing, annotation tools, and proprietary AI models, providing an end-to-end environment for cancer research and computational pathology workflows.

HistAI positions itself as a significant open-source contributor in the pathology space, having published over 112,000 WSIs on Hugging Face, released two foundation models for pathology under the Apache 2.0 license, and made available over 5.3 million manually labelled images. The platform is used by over 3,000 researchers at leading organizations.

Data Hub

• Provides access to a curated library of whole slide images, with the total dataset exceeding 212,000 slides, including over 48,333 IHC slides.
• Supports custom cohort creation with filters for tissue type, stain, clone, and disease, covering a range of diagnoses such as adenocarcinoma, ductal carcinoma, glioblastoma, melanoma, and others across multiple organs.
• Stain types available include H&E, IHC, and specific markers such as CD31, S-100, GFAP, Ki67, CD117, and others.
• Cohort statistical distributions can be visualized before purchase.
• All WSIs are available with perpetual commercial licenses suitable for research and development.
• Pricing is tiered by volume and slide type: H&E slides range from $5.00 per slide (1–99) down to $2.00 (10,000+); other slide types range from $40.00 (1–99) down to $23.00 (10,000+).
• Cohorts are formed and available for download within minutes of payment.
• Data Hub Skills toolset supports autonomous access to over 160,000 WSIs and is compatible with Claude Code, Gemini, and CODEX agents.

Slide Viewer and Annotation Tools

• Cloud-based slide viewer supports exploration at standard magnification levels (e.g., 20x).
• Provides a variety of manual annotation tools for marking and labelling slides.
• Supports real-time multi-user collaboration, including simultaneous annotation by multiple users.
• Annotations (both manual and automated) can be exported as JSON files in the Business tier.

AI Models and Analysis

• Includes a growing library of proprietary, specialized pathology AI models for slide analysis.
• AI analysis workflows provide outputs such as diagnosis detection with confidence scores (e.g., ductal cell carcinoma detected at 94.2% confidence).
• Foundation models — the Hibou family of vision transformers — were pretrained on over 1 million whole slide images and are available as open-source releases.
• Supports multi-cohort biomarker studies, with the ability to process and compare patient groups simultaneously.
• Business tier includes custom AI model development.

Platform Capabilities by User Type

• Pathologists: Access to AI-assisted analysis workflows and intuitive annotation tools intended to reduce manual review time.
• Researchers: Access to large, diverse WSI datasets to support computational pathology and discovery research.
• Pharma and Biotech: Supports biomarker development using AI models trained on diverse patient cohorts, with multi-cohort study functionality.

Pricing Tiers

• Personal: $39/month (billed annually); includes 10 GB storage, 20 AI widget runs, access to the Data Hub (100,000+ WSIs), proprietary AI models, real-time multi-user collaboration, advanced WSI labelling tools, and add-on packs for storage and AI widget runs.
• Business: Custom pricing via sales contact; includes unlimited storage and AI widget runs, unlimited seats, task management and tracking, role-based access, API access, shared workspace, preferred Data Hub pricing, GPU queue priority, custom AI model development, and custom integrations.

CellDX is hosted on Microsoft Azure, providing cloud-based scalable storage. The platform integrates with external AI agent frameworks (Claude Code, Gemini, CODEX) via the Data Hub Skills toolset. HistAI has also partnered with Protege to distribute pathology datasets to AI developers, and publishes research through peer-reviewed and preprint channels, including datasets such as HISTAI and SPIDER.