A validated active contour method driven by parabolic arc model for detection and segmentation of mitochondria

Area of Science:

• Cellular Biology
• Neuroscience
• Biomedical Imaging

Background:

• Mitochondria play a critical role in cellular functions linked to neurodegenerative aging diseases.
• Mitochondrial membrane and crista morphology are crucial for understanding mitochondrial function.
• Electron microscope tomography (EMT) provides detailed 3D structural data of mitochondria.

Purpose of the Study:

• To improve and extend automated methods for detecting and segmenting mitochondria from transmission electron microscopy (TEM) images.
• To accelerate the study of mitochondrial structure/function relationships by minimizing manual segmentation effort.
• To present a 3D extension of the developed segmentation algorithm.

Main Methods:

• Utilized a parabolic arc model for extracting mitochondrial membrane structures.
• Employed curve energy-based active contours for initial mitochondrial region outlining.
• Implemented a validation process for final segmentation data refinement.

Main Results:

• Achieved an average F-score performance of 0.84 for mitochondria detection and segmentation.
• Obtained an average Dice Similarity Coefficient of 0.87.
• Measured an average boundary error of 14nm.

Conclusions:

• The improved automated method effectively segments mitochondria in TEM images.
• The algorithm facilitates accelerated research into mitochondrial structure-function relationships.
• The 3D extension enables comprehensive analysis of mitochondrial morphology and its implications in aging and disease.