From a foundation model to applications

DeepPathAI brings pathology images, case information and professional knowledge together. It supports the development of specialist analysis, pathology Q&A and reporting assistance, with functions configured for the specialty, software version and project.

Model and application relationships
Component	Role	Scope for evaluation
DeepPathAI	Multimodal foundation for images, case context and knowledge.	Agree the model/application version and specialty task.
D-PathAI	Image analysis, visual annotations and quantitative results for review.	Confirm the selected module, supported input and intended use.
D-PathPIS	Case information, pathology Q&A and reviewable report assistance within the workflow.	The pathologist reviews suggestions and confirms the final report.
DiAssistant	DeepPathAI-powered knowledge/literature retrieval, voice interaction, content tools and report drafts.	Available with D-CleverEye DI-MSA-9; confirm the project configuration.

Explore an image-analysis module · Explore DiAssistant in microscopy

Research context

CPath-Omni · CVPR 2025

CPath-Omni unifies classification, image description, visual Q&A and region-specific prompting across image patches and whole slides. The research helps explain the capabilities being developed for flexible pathology applications.

CPath-Omni Figure 1 showing patch and whole-slide research tasks — Sun et al. · CPath-Omni, CVPR 2025 · Figure 1. Research framework and task examples.

Read the original CPath-Omni paper →

PathMMU · ECCV 2024

PathMMU evaluates pathology understanding through image-based questions and explanations. Pathologists reviewed validation and test questions. It provides a reference for professional understanding, reasoning and robustness, rather than a clinical diagnostic-performance claim.

PathMMU Figure 1 showing data sources, expert-reviewed questions and research evaluation — Sun et al. · PathMMU, ECCV 2024 · Figure 1. Research evaluation.

Read the original PathMMU paper →

These are research figures, not screenshots of a released product. Research tasks and benchmark results do not automatically establish the functions or clinical performance of a particular deployed software version.

Assistance within case review

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.

Provide relevant case context.
Review the assistant’s response or draft.
Confirm the final interpretation and report.

Original D-PathPIS assistant screen showing pathology workflow Q&A — Original product screen. Functions depend on the specialty, version and project.

Agree a specialty evaluation

Define the task: Agree the specialty use case, available data and evaluation scope.
Adapt the model: Review specialty images, local knowledge and reporting conventions.
Connect the application: Integrate the agreed image-review, reporting or quality workflow.
Evaluate together: Review with pathologists and agree the deployment scope and software configuration.

For an on-premises project, review server and storage requirements, access and update responsibilities. For cloud deployment, confirm the hosting location, access, retention, backup and contractual arrangements. Scanner and LIS integration requires a review of the actual models, file formats and interfaces.

Review deployment planning →

Plan the next step with your team

Share your use case and current setup so the team can confirm the configuration, supporting information and evaluation scope.

Discuss your configuration Back to the overview

Public content reviewed: 25 October 2026. Software and configuration versions are confirmed for the proposed project.