Using content-based image retrieval of dermoscopic images for interpretation and education: A pilot study

Area of Science:

• Dermatology
• Medical Informatics
• Computer Vision

Background:

• Dermoscopic content-based image retrieval (CBIR) systems aid in diagnosing skin lesions by providing similar images.
• Current understanding of user interaction with CBIR systems and their educational utility is limited.

Purpose of the Study:

• To develop an interactive CBIR interface for dermoscopic images.
• To investigate user interaction and decision-making with the CBIR system.
• To assess the educational value and usability of the CBIR interface.

Main Methods:

• Developed an interactive user interface for a CBIR system.
• Conducted a pilot experiment with 14 non-medically trained users.
• Evaluated user performance and feedback using CBIR and non-CBIR interfaces.

Main Results:

• Users showed increased correct classifications and confidence levels with the CBIR interface.
• User engagement and perceived ease of use were high.
• A significant increase in classification time was observed with the CBIR interface.

Conclusions:

• The CBIR system enhanced diagnostic accuracy and user confidence.
• The CBIR interface demonstrated significant educational value for learning about skin conditions.
• The system served as a useful decision support tool for image interpretation.