Sign in Subscribe
Machine learning

New Algorithm for Automatic Skin Ulcer Detection Shows Promise

Strategic Insights

— 1 min read

A team of researchers from the University of Bejaia, Algeria, has developed a new algorithm for detecting skin cancerous ulcers using image processing and machine learning techniques. The proposed method, dubbed DullRazor, aims to improve diagnostic accuracy by enhancing image quality and extracting key features from ulcer images. According to the research, the algorithm has shown promising results in detecting skin ulcers, with a dermatologist's subjective evaluation determining whether the detection is "Good," "Fairly good," or "not detected."

Key Takeaways:

  • The DullRazor algorithm is a two-phase method consisting of learning and detection phases, preceded by a crucial pre-processing step to enhance image quality.
  • The algorithm uses a dataset of 200 manually annotated ulcer images to identify distinguishing characteristics, including texture, relative color, and color features.
  • The extracted features are compared against reference vectors, and the region is classified as ulcerated if the features closely match.
  • The algorithm has been evaluated using subjective and objective methods, with promising results in detecting skin ulcers.
  • The research has been peer-reviewed and published in the journal Traitement du Signal.
  • The DullRazor algorithm has the potential to improve diagnostic accuracy in skin diseases and conditions.

Statistics:

  • The algorithm uses a dataset of 200 manually annotated ulcer images.
  • The algorithm extracts three key reference feature vectors: Texture, Relative Color, and Color features.
  • The extracted features are compared against reference vectors using an analysis window that scans the lesion.
  • The algorithm has been evaluated using subjective and objective methods.
  • The research has been peer-reviewed and published in the journal Traitement du Signal (2025;42(4):2077-2086).

Sources:

  • A New Algorithm for the Automatic Skin Ulcer Detection Using Color Features. Traitement du Signal, 2025;42(4):2077-2086.
  • NewsRx. New Data from University of Bejaia Illuminate Findings in Skin Ulcers (A New Algorithm for the Automatic Skin Ulcer Detection Using Color Features). Journal of Engineering. October 20, 2025; p 1623.

Read more