Classification of non-tumorous skin pigmentation disorders using voting based probabilistic linear discriminant...

Area of Science:

• Dermatology
• Computer Science
• Medical Imaging

Background:

• Non-tumorous skin pigmentation disorders significantly impact patients' emotional well-being.
• Accurate diagnosis is crucial for effective treatment of these conditions.

Purpose of the Study:

• To develop and evaluate a computerized method for classifying five non-tumorous skin pigmentation disorders.
• To address the challenge of large within-class variance in pigmentation images.

Main Methods:

• Probabilistic linear discriminant analysis (PLDA) was employed as the core classification technique.
• A novel voting-based PLDA (V-PLDA) approach was proposed to enhance classification accuracy.
• The V-PLDA method was tested on a dataset of 150 real-world patient images.

Main Results:

• The V-PLDA method achieved significantly higher classification accuracy compared to the original PLDA.
• Accuracy improvements exceeded 4% (p<0.001 via ANOVA) over PLDA and other state-of-the-art methods.
• This study represents the first focus on non-tumorous skin pigmentation image classification.

Conclusions:

• The V-PLDA method offers a promising solution for accurate classification of non-tumorous skin pigmentation disorders.
• This research establishes a benchmark for future studies in this domain.
• The proposed method shows potential for clinical applications in diagnosing skin pigmentation conditions.