Automated tissue classification framework for reproducible chronic wound assessment.

Area of Science:

• Medical Image Analysis
• Computational Pathology
• Wound Healing Research

Background:

• Chronic wounds (CW) pose a significant clinical challenge, requiring accurate tissue assessment for effective treatment.
• Objective evaluation of wound bed tissue composition (granulation, necrotic, slough) is crucial for healing prognosis.
• Current methods for wound tissue classification can be subjective and time-consuming.

Purpose of the Study:

• To develop and validate a computer-assisted scheme for automated tissue classification in chronic wounds.
• To utilize medical image processing and statistical machine learning for objective wound evaluation.
• To enhance the accuracy and efficiency of chronic wound assessment.

Main Methods:

• Red-green-blue (RGB) wound images were converted to the Hue-Saturation-Intensity (HSI) color space, selecting the 'S' component for enhanced contrast.
• Fuzzy divergence-based thresholding was employed for segmenting wound areas, minimizing edge ambiguity.
• Color and textural features were extracted from segmented tissues, and Bayesian classification and Support Vector Machine (SVM) algorithms were trained and tested.

Main Results:

• The Support Vector Machine (SVM) algorithm, particularly with a 3rd order polynomial kernel, demonstrated high classification accuracies: 86.94% for granulation, 90.47% for slough, and 75.53% for necrotic tissue.
• The automated tissue classification technique achieved an overall accuracy of 87.61%.
• The highest kappa statistic value of 0.793 indicated strong agreement and reliability of the proposed method.

Conclusions:

• The developed computer-assisted scheme provides an accurate and automated approach for classifying granulation, slough, and necrotic tissues in chronic wounds.
• The integration of medical image processing and machine learning offers a promising tool for objective wound evaluation.
• This technique has the potential to improve clinical decision-making and patient outcomes in chronic wound management.