A morphometric analysis of the osteocyte canaliculus using applied automatic semantic segmentation by machine...

Area of Science:

• Bone Biology
• Cellular Mechanobiology
• Biomedical Imaging

Background:

• Osteocytes function as mechanosensors, detecting mechanical stimuli via their processes.
• High-resolution imaging is crucial for analyzing osteocyte mechanosensing.
• Focused ion beam-scanning electron microscopy (FIB-SEM) offers nanometer-scale visualization.

Purpose of the Study:

• To apply machine learning for automatic segmentation of osteocyte processes and the canalicular wall.
• To perform morphometric analysis on 3D-reconstructed images of these structures.

Main Methods:

• Utilized FIB-SEM to acquire serial-section images of mouse femur osteocyte processes and canaliculi at 2 nm/voxel resolution.
• Employed machine learning for automated semantic segmentation of cellular structures.
• Validated segmentation accuracy using the Dice Similarity Coefficient (DSC).
• Reconstructed 3D images from segmented data for morphometric analysis.

Main Results:

• Achieved a DSC greater than 83% for semantic segmentation accuracy.
• Reconstructed a 3.5 μm long 3D image of osteocyte canaliculi.
• Determined the median osteocyte process diameter to be 73.8 ± 18.0 nm.
• Found the median pericellular fluid space around osteocyte processes to be 40.0 ± 17.5 nm.

Conclusions:

• Successfully implemented machine learning for semantic segmentation of osteocyte processes and canalicular walls.
• Enabled the first 3D morphometric analysis of these bone microstructures using this technique.