Clinical Skin Photo (rash or lesion)

What it does

Clinical Skin Photo reads the everyday case: one to six phone or camera photos of a rash or a skin lesion, with a short history. The run first checks each photo for blur, lighting, framing, distance and obstruction, and says how to retake it when it cannot be graded. It then returns a top-5 differential over several hundred skin conditions, each coded in ICD-11 and SNOMED CT, with "cannot exclude" flags for conditions that should stay on the list.

An urgency class suggests the next step: urgent referral for suspected skin cancer, same-day review for a possible dermatological emergency, routine referral, or self-care. When the photos follow a scoring protocol (body regions, face views or scalp views), the run also returns the matching severity score: PASI for psoriasis, EASI for atopic dermatitis, an acne grade, VASI for vitiligo or SALT for alopecia areata.

Intended use

Decision support for the clinician who sees the skin first, often without a dermatologist: a GP, an ED doctor, a nurse, or a teledermatology service sorting incoming cases. The severity scores support dermatologists and clinical trials that track treatment response with a fixed photo protocol.

Who it is for

GPs, emergency physicians, nurses, dermatologists, teledermatology providers and clinical-trial teams. Developers can call the run from a telehealth intake form to pre-fill the differential and urgency of each case for the dermatologist.

Accepted input

• Photos: JPEG, PNG, HEIC or WebP from a phone or camera, or DICOM VL Photographic Image (XC) objects.
• One case per request: one to six photos of the same complaint; a close-up together with an overview photo is recommended.
• Scoring sets: photos of four body regions for PASI and EASI, the face from the front and both sides for acne, the affected areas (optionally under Wood's lamp) for VASI, and four standard scalp views for SALT.

Requirements

Requirement	Why
A sharp, well-lit photo of the affected skin	Photos that cannot be graded are flagged with retake advice
An overview photo as well as a close-up	Distribution across the body is part of the differential
The full photo protocol for a severity score	Each score adds up area and intensity per region; missing regions leave it blank

Optional context

Age, sex, body site, duration, symptoms such as itch or pain, and how the rash has evolved, in clinical_context. These fields change the differential materially, especially age, duration, itch and distribution.

The result

One JSON result with a ranked, coded differential[], a management class and the severity sections that the photo set supports.

Section	Content
Photo quality	Gradable or not, with reasons (blur, lighting, framing, distance, obstruction) and retake advice
Differential diagnosis	Top-5 conditions with probabilities and "cannot exclude" flags, coded in ICD-10/ICD-11 and SNOMED CT
Urgency triage	Suspected skin cancer (urgent referral), possible dermatological emergency (same day), routine or self-care
PASI (psoriasis)	Erythema, induration, desquamation and area for each of four body regions, and the PASI total
EASI (atopic dermatitis)	Sign intensities and area per region with the EASI total; objective SCORAD optionally
Acne grade	Lesion counts by type and a severity grade
VASI (vitiligo)	Depigmented area per region and the VASI or facial VASI
SALT (alopecia areata)	Percentage of scalp hair loss and the SALT score
Draft report	Opt-in (options.report): an English narrative that interprets the result for the reader (findings, impression, limitations); every number is checked against the findings

Standards

• Scores: PASI (Fredriksson & Pettersson 1978); EASI (Hanifin 2001); SCORAD (European Task Force on Atopic Dermatitis); acne IGA, GAGS and Hayashi grading; VASI (Hamzavi 2004); SALT (Olsen 2004).
• Referral: suspected-cancer referral criteria such as NICE NG12.
• Coding: ICD-10, ICD-11 chapter 14 and SNOMED CT; body site in DICOM CID 4029.
• Exports: FHIR DiagnosticReport with Observations for scores and probabilities; Encapsulated PDF.