Texture based segmentation of epithelial layer from oral histological images

Area of Science:

• Digital pathology
• Medical image analysis
• Oral cancer research

Background:

• Oral sub-mucous fibrosis (OSF) originates in the epithelial layer.
• Accurate segmentation of the epithelial layer is vital for OSF diagnosis.
• Existing segmentation methods may lack precision for this specific application.

Purpose of the Study:

• To develop and evaluate a texture-based segmentation algorithm for epithelial layer delineation.
• To improve the accuracy of distinguishing normal tissue from OSF in histological images.
• To provide clinicians with a more reliable tool for diagnostic decisions.

Main Methods:

• Computed Gabor-based texture gradients on grayscale histological images.
• Extracted color gradients in the transformed Lab color space.
• Utilized watershed segmentation, combining texture and color gradients for layer segmentation.

Main Results:

• The proposed texture-based segmentation demonstrated superior performance compared to Otsu's method.
• Lower misclassification error was observed with the new algorithm.
• Segmentation results were validated against expert-provided ground truth images.

Conclusions:

• The developed texture and color gradient-based segmentation algorithm offers enhanced accuracy for epithelial layer delineation in oral histology.
• This method shows significant potential for improving the diagnosis of oral sub-mucous fibrosis (OSF).
• The algorithm provides a valuable tool for clinicians in diagnostic decision-making.