Dermatology Image Annotation FHIR Implementation Guide - Local Development build (v0.1.0) built by the FHIR (HL7® FHIR® Standard) Build Tools. See the Directory of published versions

Background

Background

The standardization gap in dermatology imaging

Radiology has decades of investment in imaging interoperability. DICOM provides a universal format for image acquisition and storage, IHE profiles define workflows for image sharing, and PACS systems offer standardized retrieval and display. When a radiologist annotates a finding, that observation flows through well-defined channels into the patient record.

Dermatology has no equivalent infrastructure for clinical photography annotation. Clinicians capture images with phones, tablets, or dedicated cameras. Annotation platforms store results in proprietary formats. AI classification tools produce outputs that cannot be exchanged between systems. The result is a fragmented landscape where annotation data is siloed within individual tools and institutions.

No existing FHIR IG addresses this need

As of March 2026, no FHIR Implementation Guide for dermatology image annotation exists in the HL7 FHIR IG Registry. Several IGs address adjacent areas (US Core covers general observations, IHE profiles cover radiology imaging workflows), but none define the specific patterns needed for structured dermatology annotation results, including skin finding classification, annotation method tracking, QA review, and AI prediction management.

Research standards without clinical integration

The International Skin Imaging Collaboration (ISIC) has made significant contributions to dermatology imaging standards for research. ISIC defines diagnostic categories, publishes benchmark datasets, and organizes classification challenges. However, ISIC does not define a FHIR output format, and its metadata conventions do not map directly to clinical terminology systems. Published research has documented the training data standardization gap: dermatology datasets lack a standard methodology for annotation capture and have no minimum metadata requirements.

Regulatory and standards momentum

Two recent developments signal growing attention to this space. The ONC Diagnostic Imaging Interoperability Request for Information (January 2026) explicitly addresses gaps in imaging data exchange beyond radiology. The HL7 AI Transparency on FHIR IG (ballot January 2026) establishes patterns for identifying AI-generated clinical content but does not include domain-specific reference implementations.

This IG addresses both needs: it defines a domain-specific annotation model for dermatology while serving as a reference implementation of AI Transparency patterns in a clinical imaging context.