Clinical studies, publications, and supporting  evidence.

Performance of the ‘DermDx’ algorithm in triaging suspected skin cancers: results of the AI-SCSS study

This real-world UK study evaluated MetaOptima’s DermDx AI for triaging 1,024 lesions referred through a suspected skin cancer pathway. DermDx correctly identified 296 of 298 confirmed skin cancers, achieving 99.3% sensitivity for cancer and 98.1% sensitivity for malignant and premalignant lesions combined. The researchers estimated that using DermDx for triage could reduce face-to-face dermatology consultations by 26.7%, while maintaining cancer-detection performance comparable to dermatologists. 

Read the study

Perspectives and experiences of patient-led melanoma surveillance using digital technologies from clinicians in the MEL-SELF pilot RCT

Interviews with eight clinicians involved in the MoleScope/DermEngine-enabled pilot, covering workflow benefits, trust, workload and implementation concerns.

Diagnostic accuracy of content-based dermatoscopic image retrieval with deep classification features

MetaOptima funded the work and supplied hardware; 

• Content‐based image retrieval (CBIR) based on deep features can find visually similar dermatoscopic images.
• Retrieving only 16 similar images can achieve the same accuracy as a CNN classifier.
• CBIR can enable a CNN to recognize unknown disease classes in new datasets.

Using content-based image retrieval of dermoscopic images for interpretation and education: a pilot study

Small pilot with 14 non-medically trained participants evaluating an academic image-retrieval interface. Includes a MetaOptima-affiliated author but is not direct commercial-product validation.

Can self-training identify suspicious ugly duckling lesions?

Peer-reviewed computer-vision conference paper with MetaOptima-affiliated researchers. Reports 72.1% sensitivity and 94.2% diagnostic accuracy on its held-out test set — technical research, not validation of a named marketed product.

Can patient-led surveillance detect subsequent new primary or recurrent melanomas and reduce routinely scheduled follow-up? MEL-SELF RCT protocol

Protocol for the larger MEL-SELF randomized controlled trial.

Acceptability of a hypothetical reduction in routinely scheduled clinic visits among patients with a history of localized melanoma (MEL-SELF)

Pilot randomized clinical trial sub-study on patient acceptability of reduced clinic follow-up.

Participant motivators and expectations in the MEL-SELF randomized clinical trial: content analysis of survey responses

Content analysis of why participants join patient-led surveillance and what they expect from it.

Characteristics of participants screened and randomized to the Melanoma Self Surveillance trial

Secondary analysis of 504 randomized participants. Describes the intervention as including a mobile dermatoscope and teledermatologist assessment.

Experiences of patient-led melanoma surveillance and teledermatology in underserved groups: a qualitative sub-study

Implementation evidence from the larger MEL-SELF program.

Teledermatology and teledermoscopy as tools to help general practitioners in effective e-triage of suspicious skin lesions

Direct MoleScope and DermEngine e-triage study involving 23 GPs. The final poster reports 292 patients, 10 melanomas, 12 basal cell carcinomas and 10 squamous cell carcinomas, with dermatologist access reduced from 1–12 months to 14–90 days.

BT08: Assessment of three different dermoscopy imaging systems during a fast-track skin cancer clinic

Independent conference abstract comparing Heine Delta 20T, MoleScope II and Dino-Lite Edge across 23 lesions. MoleScope II scored lower on mean image quality but was comparable for diagnostic ability and was the most mobile, lowest-cost system evaluated.

AI-SCSS EADV abstract and poster — real-world performance of a commercial deep neural network algorithm in a UK suspected skin cancer pathway

Preliminary counts differ from the final paper because of later exclusions and 12-month follow-up.

One million skin checks: a feasibility study of teledermatology and AI in general practice

Appears linked to ACTRN12623000275662 (entry 12); the product connection is established through the registry, not the abstract alone.

Out-of-distribution detection for dermoscopic image classification

Technical preprint on detecting novel or out-of-distribution dermoscopic disease classes.

Prospective evidence on artificial intelligence–assisted melanoma diagnostics: a systematic review and meta-analysis

Includes the Menzies 2023 study and identifies the evaluated AI group as “MetaOptima; 7-class.” Appropriate secondary evidence, but not a separate DermDx validation.