Computational normalization of H&E-stained histological images: Progress, challenges and future potential

Area of Science:

• Digital pathology
• Computational image analysis
• Cancer diagnostics

Background:

• Hematoxylin-eosin (H&E) staining is crucial for cancer diagnosis via histological sample analysis.
• Color variations in stained samples, due to preparation and digitization, hinder automated image analysis algorithms.
• These variations can compromise the accuracy of segmentation and classification in identifying cancerous lesions.

Purpose of the Study:

• To present a comprehensive review of the state-of-the-art in computational normalization of H&E-stained histological images.
• To highlight the key contributions and limitations of existing normalization methods.
• To evaluate published normalization techniques and suggest future research directions.

Main Methods:

• Systematic literature review of computational normalization techniques for H&E-stained histological images.
• Analysis of factors contributing to color variations in histological samples.
• Evaluation of the performance of various normalization algorithms.

Main Results:

• Identified key factors influencing color variability in H&E staining.
• Detailed the strengths and weaknesses of current image normalization approaches.
• Provided an overview of the performance of different normalization methods found in the literature.

Conclusions:

• Computational normalization is essential for standardizing H&E-stained histological images.
• Effective normalization methods can significantly improve the reliability of automated cancer diagnosis systems.
• Further research is needed to develop robust and universally applicable normalization techniques.