Border detection on digitized skin tumor images.

Area of Science:

• Dermatology
• Medical Imaging
• Computer Vision

Background:

• Accurate detection of skin tumor borders is crucial for diagnosis and treatment planning in dermatology.
• Existing imaging techniques may face challenges in precisely delineating tumor margins due to variations in skin texture and lesion morphology.

Purpose of the Study:

• To develop and evaluate a novel radial search algorithm for automated detection of skin tumor borders in clinical dermatology images.
• To improve the accuracy and reliability of tumor border identification compared to existing methods.

Main Methods:

• A two-round radial search technique initiated from a common tumor center.
• The first round involves an independent search, followed by a knowledge-based tracking search in the second round.
• A rescan strategy with a new center is implemented to address potential blind spots.

Main Results:

• The algorithm demonstrated excellent performance on model images.
• Satisfactory results were achieved when tested on 300 real clinical dermatology images.
• The rescan approach effectively mitigated the blind-spot problem inherent in single-center radial searches.

Conclusions:

• The presented radial search technique offers a promising approach for precise skin tumor border detection in clinical settings.
• This automated method has the potential to aid dermatologists in more accurate lesion assessment and management.
• Further validation on larger datasets could establish this technique as a valuable tool in dermatological image analysis.