Mitochondria morphometry in 3D datasets obtained from mouse brains with serial block-face scanning electron...

Area of Science:

• Neuroscience
• Cell Biology
• Biotechnology

Background:

• Mitochondrial dysfunction is linked to various diseases, particularly neurodegenerative disorders.
• Assessing mitochondrial health involves analyzing cellular metabolism, morphology, and subcellular localization.
• Electron microscopy is crucial for precise neuronal mitochondria analysis, but 3D reconstruction and segmentation are challenging.

Purpose of the Study:

• To develop and validate an open-access workflow for 3D imaging, segmentation, and analysis of mitochondria in mouse brain tissue.
• To leverage recent advancements in pre-trained models to accelerate mitochondrial reconstruction and analysis.
• To critically evaluate the impact of different reconstruction methods and manual correction levels on analysis outcomes.

Main Methods:

• Utilized serial block face scanning electron microscopy (SBF-SEM) for 3D ultrastructural imaging of mouse brain volumes.
• Implemented an open-access workflow for large-scale segmentation of neuropil and mitochondria.
• Employed pre-trained deep learning models to expedite the 3D reconstruction and analysis process.

Main Results:

• Successfully established a reproducible workflow for imaging and analyzing mitochondria in 3D ultrastructural volumes of the mouse brain.
• Demonstrated accelerated reconstruction and analysis times through the use of pre-trained models.
• Provided a critical assessment of reconstruction methods and manual correction's influence on quantitative mitochondrial analysis.

Conclusions:

• The developed open-access workflow enables efficient and detailed 3D analysis of neuronal mitochondria.
• This approach facilitates the study of mitochondrial alterations in neurodegenerative diseases and other conditions.
• The workflow's flexibility and reliance on open resources promote wider adoption in neuroscience research.