High-Throughput, Label-Free and Slide-Free Histological Imaging by Computational Microscopy and Unsupervised Learning

Area of Science:

• Biomedical Imaging
• Optical Microscopy
• Computational Pathology

Background:

• Histological imaging traditionally requires extensive tissue preparation, including fixation, sectioning, and staining, which is time-consuming and can introduce artifacts.
• Rapid, high-resolution imaging of unprocessed tissues with minimal preparation remains a significant challenge in medical diagnostics.

Purpose of the Study:

• To introduce a novel computational microscopy technique for high-throughput, label-free histological imaging.
• To enable rapid, high-resolution imaging of thick, unprocessed tissues with complex surface topography.
• To develop a method for transforming autofluorescence images into virtually stained histological images for quantitative analysis.

Main Methods:

• Computational high-throughput autofluorescence microscopy by pattern illumination (CHAMP) was employed for label-free imaging.
• Unsupervised learning (Deep-CHAMP) was utilized to transform autofluorescence images into virtually stained histological images.
• Imaging speed, lateral resolution, and quantitative feature extraction accuracy were assessed.

Main Results:

• CHAMP achieved high-throughput imaging at 10 mm²/10 s with 1.1-µm lateral resolution on thick, unprocessed tissues.
• Deep-CHAMP generated virtually stained images within 15 s, enabling accurate quantitative extraction of cellular features.
• The method demonstrated versatility across various tissue types (mouse brain/kidney, human lung) and clinical protocols.

Conclusions:

• CHAMP provides a rapid, label-free, and high-resolution histological imaging solution, significantly reducing or eliminating tissue processing time.
• The ability to generate virtually stained images and extract quantitative features enhances diagnostic accuracy for intraoperative and postoperative pathological examination.
• CHAMP shows great potential as an assistive imaging platform for surgical and pathological decision-making.