Validity of facial skin analysis pore detection: A comparative analysis

Area of Science:

• Dermatology
• Artificial Intelligence
• Medical Imaging

Background:

• Objective assessment of facial characteristics is crucial for evaluating dermatological treatments.
• Previous AI tools showed limitations in pore assessment compared to physician and computer-based evaluations.
• An updated AI algorithm specifically targeted improved pore detection.

Purpose of the Study:

• To evaluate the performance of an updated AI tool in assessing pore scores across diverse Fitzpatrick skin phototypes.
• To determine if the AI tool's pore assessments align with dermatologists' evaluations.
• To compare the updated AI tool against its previous version and a gold-standard device.

Main Methods:

• Participants' frontal photographs were captured under uniform lighting.
• Four self-learning models within the updated AI tool generated pore percentile scores.
• Pore ratings from the AI tools were compared against a gold-standard device and a board-certified dermatologist's assessment.

Main Results:

• The updated AI tool demonstrated improved pore detection accuracy across all Fitzpatrick skin types.
• Specific AI models (A and D) showed high concordance with physician assessments for Fitzpatrick skin types III-IV and V-VI, respectively.
• The AI tool's pore detection performance surpassed its original version and a computer-based program.

Conclusions:

• The AI tool's pore detection algorithm update successfully improved accuracy for all Fitzpatrick skin types.
• AI tools can reliably detect skin metrics across diverse populations, aiding dermatologic evaluation and treatment.
• Accurate AI-driven skin assessment facilitates personalized treatment plans and monitoring of therapeutic response.