Development of an image processing software for quantification of histological calcification staining images

Area of Science:

• Biomedical research
• Digital pathology
• Image analysis

Background:

• Histological staining images are crucial for biomedical research.
• Existing quantification tools struggle with noisy or unwanted stains, assuming clean backgrounds.
• Complex staining scenarios, including counterstaining and dirty stains, pose challenges for current methods.

Purpose of the Study:

• Develop a Staining Quantification (SQ) tool for accurate histological image analysis.
• Create software capable of removing unwanted stains blended with the Region of Interest (ROI).
• Address limitations of current tools in handling complex and noisy histological images.

Main Methods:

• Developed a light software tool named Staining Quantification (SQ).
• Employed higher order statistics transformation and local density filtering as core algorithms.
• Validated the tool using in-house histological images (Alizarin Red and Von Kossa) and external datasets.

Main Results:

• The SQ tool demonstrated superior performance compared to Otsu's and Triclass thresholding methods.
• Achieved an average mean difference below 0.05% in area and intensity measurements compared to ImageJ.
• Exhibited a success rate above 0.8 for measurements across various staining types and validation batches.

Conclusions:

• The Staining Quantification (SQ) tool is a robust solution for automatic histological staining image quantification.
• The software effectively handles complex scenarios with unwanted stains and weak signals.
• SQ provides accurate and reliable measurements, outperforming established methods in challenging conditions.