Home Products Solutions Technology Publications About Discuss your workflow
English Français Deutsch Español Italiano Português Bahasa Indonesia 中文 繁體中文（香港） Русский 日本語 한국어 العربية Tiếng Việt ไทย

DeepPathAI

A multimodal foundation for pathology image analysis, professional Q&A and workflow applications.

PATHOLOGY FOUNDATION MODEL

From image understanding to everyday pathology

DeepPathAI brings pathology images, case information and professional knowledge together. It works with D-PathAI and D-PathPIS to support specialist analysis, pathology Q&A and reporting assistance, and provides a foundation for joint development with hospitals and industry partners.

Explore the applicationsHospital adaptation
Original CPath-Omni figure showing patch and whole-slide inputs, classification, image descriptions, region prompts and visual question answering
Sun et al. · CPath-Omni, CVPR 2025 · Figure 1. Research framework and task examples.

CPath-Omni CVPR 2025

One framework for image patches and whole slides

CPath-Omni brings classification, image description, visual Q&A and region-specific prompting into a unified research framework. These capabilities underpin the development of flexible pathology applications across image scales.

Visual representations
Learn tissue architecture and cell morphology to support specialist classification, retrieval and image-analysis development.
Image–text alignment
Connect pathology images with professional descriptions, providing a foundation for image–text retrieval and multimodal application development.
Multimodal interaction
Combine images, case context and text instructions for image Q&A, information synthesis and reporting assistance.
View the original paper

PathMMU ECCV 2024

Evaluate understanding with pathology expertise

PathMMU uses image-based questions and answer explanations, with validation and test questions reviewed by pathologists. It examines professional understanding, reasoning and robustness to image changes, providing a reference for model evaluation and specialty adaptation.

View the original paper
Original PathMMU overview of pathology data sources, expert-reviewed multimodal questions and model evaluation
Sun et al. · PathMMU, ECCV 2024 · Figure 1. Data sources, question design and research evaluation.

PathMMU is a research benchmark. Its scores do not establish clinical diagnostic performance.

D-PathPIS · MODEL IN THE WORKFLOW

Pathology assistance within the case workflow

Bring case information and a preliminary diagnosis into pathology Q&A, report drafting and content review. The pathologist reviews the suggestions and confirms the final report within the existing workflow.

  1. Case context
  2. AI assistance
  3. Pathologist confirmation

DeepPathAI

D-PathAI Image analysisD-PathPIS Cases and reporting

Functions are configured for the specialty, software version and project.

Original D-PathPIS assistant screen showing a response about the pathology workflow
Pathology workflow Q&A in the D-PathPIS assistant. Original product screen.
A clinician reviewing a digital tissue image beside a DiPath scanner
Digital pathology review with a DiPath scanner and image workstation.

HOSPITALS AND INDUSTRY PARTNERS

Build around a defined specialty need

Develop a specialty model or integrate AI into an existing product. We work with partners on model adaptation, local knowledge and workflow integration, with a jointly agreed evaluation and rollout scope.

Deployment and integration

  1. Define the task

    Agree the specialty use case, available data and evaluation scope.

  2. Adapt the model

    Adapt to specialty images, local knowledge and reporting conventions.

  3. Connect applications

    Integrate with agreed image-review, reporting or quality workflows.

  4. Evaluate together

    Review with pathologists and confirm the deployment scope.

D-PathAI in practice

Explore AI-assisted slide review

Explore the product screen, then select a local field to inspect the image and analysis findings.

D-PathAI cervical cytology product overview showing AI candidate markers and whole-slide analysis results.

Cervical cytology

Review the product overview and original cell fields with AI candidate markers.

D-PathAI stomach H&E product overview showing the tissue heatmap and source classification.

Stomach H&E

Review the tissue heatmap and source classification in the product overview.

D-PathAI HER2 IHC product overview showing membrane-staining overlays and cell counts for the analysed region.

HER2 IHC

Inspect membrane-staining overlays and cell counts for the analysed region in this product example.

Supported analysis and intended use depend on the module, version and market. Outputs require review by a qualified user.

Discuss a module demonstration
Plan deployment with your laboratory and IT team

Start with your scanner, image formats, LIS and data-handling requirements. The final configuration depends on the agreed product and deployment.

On-premises deployment
Review server and storage requirements, user access, software updates and responsibilities with your IT team.
Cloud deployment
Confirm the hosting location, data access, retention, backup and contractual arrangements for your project.
Scanner and LIS integration
Share scanner models, file formats and LIS interfaces so the team can assess compatibility and the work needed.

Image detail

—

Loading image…