Real-time three-dimensional histology-like imaging by label-free nonlinear optical microscopy

Area of Science:

• Biomedical Optics
• Advanced Microscopy Techniques
• Computational Pathology

Background:

• Traditional histology methods (FFPE) are time-consuming and can distort delicate tissue structures.
• Current techniques require extensive sample preparation, including fixing, embedding, sectioning, and staining.
• These processes limit the ability to study dynamic in vivo tissue processes and 3D architecture.

Purpose of the Study:

• To develop a rapid, label-free imaging technique for visualizing tissue microstructure.
• To create histology-like images from fresh tissue samples without staining.
• To enable real-time 3D imaging of tissue for potential diagnostic applications.

Main Methods:

• Utilized a Simultaneous Label-free Autofluorescence Multiharmonic (SLAM) microscope integrating SHG, 2PF, THG, and 3PF modalities.
• Performed label-free imaging on fresh human and rat tissues, both ex vivo and in vivo.
• Employed deep learning for nuclei segmentation and a color-transforming algorithm to generate histology-like images.

Main Results:

• SLAM microscopy effectively visualized key tissue features like collagen and nuclei without staining.
• Generated histology-like images with high similarity to traditional H&E staining.
• Achieved real-time 3D histology-like imaging and quantified nuclear-to-cytoplasmic ratios.

Conclusions:

• SLAM microscopy offers a powerful platform for real-time 3D histology-like imaging.
• The integration of deep learning and color remapping enhances visualization capabilities.
• This technique holds significant potential for advancing diagnostic pathology and biological research.