Dermi
Clinical imaging with research discipline.
Company
Dermi Inc.
Location
Toronto, Canada
Primary product
Dermi Atlas platform
Started in research, built for practice
Dermi Atlas grew out of applied dermatology imaging work, then became the platform layer those tools needed.
Dermi began as a collection of independent research projects in dermatological image analysis. Segmentation models for estimating affected skin area, classification methods for organizing dermatological imagery, and registration algorithms for aligning images across visits were each developed and evaluated separately as academic research.
Each tool addressed a real clinical need, but none had the infrastructure to be useful in practice. There was no patient management, no image organization, no audit trail, and no way to deploy these tools in a clinical environment without significant technical overhead.
Integrating directly into existing EMR systems as plugins was explored but proved impractical for the research tools themselves. Closed-source platforms required revalidation cycles that were incompatible with the pace of active research.
- 01
Research Tools
Independent research projects for segmentation, classification, and registration, each solving a narrow problem in isolation.
- 02
Integration Gap
Integrating research tools as EMR plugins required revalidation for every update, making that approach impractical at research pace.
- 03
Platform Decision
A dedicated platform was needed to provide the infrastructure, ecosystem, and clinical features that standalone tools lacked.
The result is a dedicated clinical photography platform, not a research demo and not an EMR plugin.
Local-first by design, not by default
Clinical imaging involves sensitive patient data where the consequences of exposure extend beyond the technical.
Cloud infrastructure is the industry standard for SaaS platforms and offers real convenience. For clinical imaging, that convenience trades against the cost of a breach. Building local-first was a deliberate decision, not a limitation.
As AI and machine learning capabilities are integrated over time, this distinction becomes even more significant. Cloud-based processing would require patient images to transit external networks and be processed on shared infrastructure. Local-first architecture is designed to keep data and computation within the clinical network, so current and planned processing can operate without routing patient images through external infrastructure.
This approach asks more of deployment, and it places responsibility for the surrounding network, devices, and backups with the practice. Dermi Atlas Manager exists to make that practical: it guides self-hosted setup, updates, and backups for practices without dedicated IT staff.
Privacy model
Data Sovereignty
Patient data stays on infrastructure the practice owns and controls. By default no external servers sit in the data path and no third-party access is built in, so the practice decides who can reach its data.
Local Processing
All processing runs on local hardware, so sensitive data stays within the clinical network by design. The architecture is built to keep current and future processing on clinic-controlled infrastructure rather than shared external systems.
Architectural Privacy
Privacy is enforced by system design, not by policy alone. Keeping patient data and processing on clinic-controlled infrastructure closes off external exposure paths that a contract can only promise to avoid.
What was built and what it solves
The product system spans three parts: a clinical workspace for daily practice, a control layer for self-hosted deployment, and a native iOS app for reference-aligned capture at the point of care.
Dermi Atlas Professional
The core of the platform and the workspace clinicians use every day. Atlas Professional organizes, compares, and reports on patient imaging in one place, through an interface built for both touch and desktop. It runs on infrastructure the practice controls.
- Chronological patient image organization
- Side-by-side and overlay comparison tools
- Configurable tagging by condition and body location
- PDF export for referrals and patient records
Dermi Atlas Manager
Desktop application that makes self-hosted deployment accessible without IT staff. Handles Docker orchestration, database management, backups, and updates through a guided interface on macOS, Windows, and Linux.
- One-click installation on local hardware
- Docker container orchestration and monitoring
- Backup creation, restoration, and scheduling
- Version management and guided updates
Dermi Atlas Companion
Native iOS app for iPhone and iPad that brings the full Atlas workspace to the point of care, with native capture. Follow-up photography happens in the room and flows into the patient record, free on the App Store with no in-app purchases.
- Reference-aligned capture matched to a prior visit with live overlays
- On-device assists for angle, framing, and distance
- Multi-server access across locations with single-tap switching
- HTTPS connections pinned to the server you paired
Transparent commercial model
Straightforward per-license pricing with no hidden fees, a thirty-day free trial of Atlas Professional, and Atlas Companion free on the App Store.
Peer-reviewed and presented research informing the platform
Peer-reviewed journal studies and presented research established the methods that inform Dermi Atlas and its long-term roadmap. These research capabilities are not yet available in the current product.
Research archive
01
2022 / JAAD
Peer-reviewed journal
BSA Estimation Study
Research study on a convolutional neural network for automated psoriasis body surface area estimation, validated against dermatologist assessment with roughly one-third the error.
8.95%
mean error (vs. 27.84% dermatologist)
02
2024 / EADV Congress
Congress presentation
SPREAD Framework
SPREAD (Skin Patch-based Regional Extent Assessment for Dermatoses) is a multi-stage research framework combining segmentation, classification, and metric computation across full-body imaging.
Multi-stage
full-body assessment method
03
2024 / JEADV
Peer-reviewed journal
AI PASS Study
Research study comparing AI-based psoriasis body surface area estimation against board-certified dermatologists, with substantially lower error than expert assessment.
0.668
MAE (vs. 2.31-5.55 dermatologist)
A platform designed for extensibility
The long-term goal is an ecosystem where research translates into practice without forcing clinics to adopt separate software for every new method.
Dermi Atlas was built as a platform, not a single-purpose tool, so its architecture is intended to support capabilities beyond basic image management. Over time, the platform is intended to serve as a foundation for advanced clinical tools that would not integrate into clinical workflows if built as standalone applications.
New models and methods should be able to slot into the platform without requiring practices to replace the surrounding workflow, patient structure, deployment model, or auditability.
Advanced imaging features
Automatic image alignment is part of Atlas today. Background removal and full-body mapping are on the roadmap. Each requires a purpose-built platform and would not integrate cleanly into general-purpose EMR systems.
Open pipeline architecture
An extensible processing pipeline where research teams can integrate their own models and algorithms. They should be able to deploy new methods without rebuilding the surrounding infrastructure.
Research without rebuilding
Researchers should focus on a single processing step, not an entire system. The platform provides patient management, image storage, comparison tools, and deployment, so research can target the pipeline, not the product.
Research should target the processing pipeline, while the clinical platform handles the surrounding system.
Principles that drive decisions
Core values reflected in architecture, feature scope, documentation, and business practice.
Data Sovereignty
Patient data belongs to the practice. A self-hosted architecture keeps clinical data on clinic-controlled infrastructure and supports HIPAA, PIPEDA, and Australian Privacy Act requirements. Each practice remains responsible for its own compliance.
Accessibility
Self-serve deployment that does not require IT staff or specialist setup. Interfaces built for both touch and desktop, plus a native iOS app for capture at the point of care. Technology that meets clinicians where they already are.
Transparency
Open pricing with no hidden fees, and a free trial before any commitment. Clear, current documentation for every feature. Honest communication about what the platform does today, what it does not, and what is still on the roadmap.
Learn more about Dermi
Explore the newsroom, contact the team, review policies, or evaluate Dermi Atlas directly.
Experience Dermi Atlas
Try the Cloud Demo in your browser, or start a free trial and run Dermi Atlas on your own hardware.