Automated basal cell carcinoma detection in high-definition optical coherence tomography

Area of Science:

• Dermatology
• Medical Imaging
• Artificial Intelligence

Background:

• Basal cell carcinoma (BCC) is the most prevalent non-melanoma skin cancer.
• Current BCC diagnosis relies on invasive skin biopsies.
• High-definition optical coherence tomography (HD-OCT) offers non-invasive in vivo skin imaging.

Purpose of the Study:

• To develop and validate an automated method for detecting basal cell carcinoma (BCC) using HD-OCT imaging.
• To establish the feasibility of using HD-OCT for non-invasive BCC screening.

Main Methods:

• A novel four-step BCC detection method was proposed: graph-based skin surface segmentation, surface flattening, deep feature extraction, and BCC classification.
• The method was evaluated on a dataset comprising 5,040 HD-OCT images.

Main Results:

• The proposed method demonstrated effectiveness in automatically detecting BCC from HD-OCT images.
• Strong agreement between HD-OCT imaging features and histopathological findings was observed.

Conclusions:

• The developed automated BCC detection method using HD-OCT shows significant potential.
• This technique can serve as an effective, non-invasive tool for screening basal cell carcinoma, potentially reducing the need for biopsies.