Using Artificial Intelligence to Automate the Analysis of Psoriasis Severity: A Pilot Study

Area of Science:

• Dermatology
• Artificial Intelligence
• Medical Imaging

Background:

• The Psoriasis Area and Severity Index (PASI) is standard for assessing psoriasis severity.
• Manual PASI scoring faces challenges due to environmental variability and subjective interpretation.
• This study explores AI to enhance the objectivity and consistency of psoriasis severity classification.

Purpose of the Study:

• To develop and evaluate an AI model for objective psoriasis severity classification.
• To improve the consistency of PASI scoring using features from 2D clinical images.
• To classify psoriatic lesions based on erythema, thickness, and scaling severity.

Main Methods:

• Utilized the YOLOv8 deep learning model for image classification.
• Trained the model on three diverse datasets in a cloud environment (Google Colab).
• Employed stratified k-fold cross-validation for robust performance assessment.

Main Results:

• The YOLOv8 model demonstrated high effectiveness in classifying psoriasis images according to PASI scores.
• Stratified k-fold cross-validation improved model reliability across different datasets.
• The AI model accurately classified lesion severity based on key PASI subcomponents.

Conclusions:

• AI, specifically YOLOv8, offers a significant advancement for automated psoriasis severity classification.
• This AI-driven approach enhances the objectivity of assessing erythema, thickness, and scaling.
• The study highlights the potential of AI in standardizing psoriasis assessment.