Automatic quantification of IHC stain in breast TMA using colour analysis.

Area of Science:

• Pathology
• Biomedical Imaging
• Computational Biology

Background:

• Immunohistochemical (IHC) biomarkers in breast tissue microarrays (TMA) are crucial for cancer diagnosis.
• Subjective evaluation of IHC staining leads to diagnostic errors and time inefficiencies.
• Automated methods are needed to objectively quantify IHC staining in pathology.

Purpose of the Study:

• To develop an automated density tool for quantifying positive brown IHC stain in breast TMA samples.
• To create a robust tool independent of staining variations using color standardization.
• To compare automated segmentation with expert pathologist evaluations.

Main Methods:

• Analysis of four color standardization methods against color model segmentation using NBS color distance.
• Combination of HSV and RGB color models with HSB channels for robust segmentation.
• Quantitative validation using Hausdorff distance, sensitivity, and specificity on 8000 TMA images.
• Development of a tool providing outputs for weak, medium, and strong staining intensity.

Main Results:

• The developed automated tool achieved an average accuracy of 95.94% across 8000 TMA images.
• Color standardization effectively reduced noise from staining and sample preparation.
• The combined HSV/RGB and HSB approach yielded the most reliable segmentation results.
• Automated quantification showed high agreement with expert pathologist assessments.

Conclusions:

• The automated density tool accurately quantifies IHC staining in breast TMA, offering a reliable alternative to manual evaluation.
• The method enhances diagnostic efficiency and consistency in pathology departments.
• This technology has the potential to significantly reduce errors in breast cancer diagnosis.