Enhanced Multiscale Human Brain Imaging by Semi-supervised Digital Staining and Serial Sectioning Optical Coherence...

Area of Science:

• Neuroscience
• Biomedical Imaging
• Computational Biology

Background:

• Traditional histology methods for brain visualization are limited by staining variability, tissue damage, and distortion, hindering accurate 3D reconstructions.
• Accurate multiscale visualization of human brain structures is crucial for understanding neural organization and function.

Approach:

• Developed a novel 3D imaging framework integrating serial sectioning optical coherence tomography (S-OCT) with a deep-learning digital staining (DS) model.
• Employed a semi-supervised learning technique for training the DS model on weakly paired images, enabling translation from S-OCT to Gallyas silver staining.
• Demonstrated the framework's capability on human cerebral cortex samples, achieving consistent staining quality and enhanced contrast across cortical layers.

Key Points:

• The digital staining model successfully translates S-OCT images to Gallyas silver staining, offering consistent quality across samples.
• The technique enhances contrast at cortical layer boundaries, improving the delineation of brain structures.
• Geometry-preserving 3D digital staining was showcased on large tissue blocks, enabling visualization of meso-scale vessel networks in white matter.

Conclusions:

• The presented 3D imaging framework offers a powerful new tool for high-throughput, multiscale imaging of brain tissues.
• This approach has the potential to significantly facilitate studies of complex brain structures and their organization.
• The digital staining method overcomes key limitations of traditional histology, paving the way for more accurate and detailed brain mapping.