Towards data-driven quantification of skin ageing using reflectance confocal microscopy

Area of Science:

• Dermatology and biomedical engineering
• Computational analysis of biological images
• Machine learning applications in aging research

Background:

• Skin aging evaluation is subjective and lacks standardized measures.
• Reflectance confocal microscopy (RCM) provides quantitative indicators of skin aging.
• Developing an objective, data-driven method for skin age assessment is needed.

Purpose of the Study:

• To create a machine learning model for objective skin aging quantification.
• To develop a data-driven scale for skin aging using RCM image analysis.
• To eliminate human assessment subjectivity in skin aging evaluation.

Main Methods:

• Utilized RCM depth stacks from 74 participants.
• Employed unsupervised clustering and logistic regression for feature analysis.
• Trained a classifier to differentiate skin from young (20-30) and older (50-70) age groups.

Main Results:

• The classifier achieved an Area Under the Curve (AUC) of 0.908 in differentiating age groups.
• Significant differences in aging scores were observed between age groups and body sites (dorsal vs. volar).
• 17 out of 20 validation samples were correctly classified.

Conclusions:

• Machine learning effectively quantifies skin aging from RCM data.
• The model shows potential for differentiating chronological aging and sun exposure effects.
• This approach enables fine-grained, data-driven quantification of skin aging.