Dermoscopic Features of Cutaneous Vasculitis

Area of Science:

• Dermatology
• Medical Imaging
• Artificial Intelligence in Medicine

Background:

• Dermoscopy is increasingly used for diagnosing inflammatory skin conditions.
• Cutaneous vasculitis (CV) requires rapid and reliable diagnostic methods due to vessel inflammation.

Purpose of the Study:

• To define characteristic dermoscopic features of cutaneous vasculitis (CV).
• To improve the diagnostic accuracy of dermoscopy for CV using machine learning (ML) techniques.

Main Methods:

• Eighty-nine patients with suspected CV underwent polarized dermoscopy before biopsy.
• Dermoscopic images were evaluated independently, with interobserver variability assessed.
• Machine learning models including Decision Tree, Random Forest, and K-Nearest Neighbors were employed for classification.

Main Results:

• Histopathology confirmed CV in 58 patients. Observed patterns included homogeneous, mottled, and meshy.
• Specific dermoscopic markers for CV included milky red (sensitivity 56.7%, specificity 96.3%) and livedoid (sensitivity 29.4%, specificity 99.2%) background colors.
• Red blotches were more frequent in CV lesions (P=0.038), while red dots, comma vessels, and scales were more common in non-CV cases (P<0.003).

Conclusions:

• Dermoscopy serves as a rapid and reliable tool for diagnosing cutaneous vasculitis (CV).
• The integration of machine learning methods significantly enhances dermoscopic diagnostic accuracy for CV.