Robust autofocus method based on patterned active illumination and image cross-correlation analysis.

Area of Science:

• Digital pathology
• Biomedical imaging
• Computational pathology

Background:

• Whole-slide image (WSI) quality is crucial for computer-aided diagnosis (CADx) systems.
• Existing autofocus methods face challenges in balancing speed, accuracy, and adaptability across different samples and imaging scenarios.
• Accurate autofocus is essential for reliable WSI generation and subsequent analysis.

Purpose of the Study:

• To develop a robust autofocus method that overcomes the limitations of current techniques.
• To ensure high-quality WSI generation for improved CADx system performance.
• To provide a versatile autofocus solution applicable to various imaging conditions and sample types.

Main Methods:

• A novel autofocus method utilizing fiber bundle illumination and image normalization analysis.
• Active illumination strategies to accommodate diverse application scenes, including bright field and fluorescence imaging.
• Image analysis techniques to guarantee autofocusing accuracy for varying sample structures.

Main Results:

• The proposed method effectively tracks changes in sample-to-focal plane distance.
• Demonstrated ability to maintain autofocus accuracy across different sample structures and imaging modalities.
• Significant improvement in overall WSI quality was observed.

Conclusions:

• The developed autofocus method offers a robust and adaptable solution for WSI acquisition.
• It enhances the foundational quality of WSIs, thereby supporting more effective CADx systems.
• This approach addresses the need for a universal autofocus technique in digital pathology.