Color Normalization in Breast Cancer Immunohistochemistry Images Based on Sparse Stain Separation and Self-Sparse...

Area of Science:

• Digital pathology
• Medical image analysis
• Computational biology

Background:

• Immunohistochemistry (IHC) staining is crucial for breast cancer diagnosis and treatment.
• Color variations in IHC images can hinder accurate interpretation and affect the Allred score.
• Pathologists rely on consistent image quality for precise drug quantity estimation.

Purpose of the Study:

• To develop and evaluate a new color normalization technique for breast cancer IHC images.
• To enhance the interpretability of IHC-stained images for pathologists.
• To improve the accuracy of Allred score assessment.

Main Methods:

• A novel color normalization technique utilizing sparse stain separation and self-sparse fuzzy clustering.
• Quaternion structural similarity was employed to quantify normalization quality.
• Automated, unsupervised nuclei classification with Automatic Color Deconvolution (ACD) was used to assess extracted color features.

Main Results:

• The proposed technique achieved a lower structural similarity score compared to existing methods, indicating better normalization.
• The color distribution similarity of the normalized images was closer to the target.
• Unsupervised nuclei classification using ACD on normalized images yielded results comparable to other methods.

Conclusions:

• The new color normalization method enhances image interpretability for pathologists.
• The technique provides a more accessible perception of IHC-stained images.
• This approach supports more reliable Allred score determination in breast cancer treatment planning.