ECLiPSE: a versatile classification technique for structural and morphological analysis of 2D and 3D single-molecule...

Area of Science:

• Cell Biology
• Microscopy
• Bioinformatics

Background:

• Single-molecule localization microscopy (SMLM) provides nanoscale resolution for visualizing subcellular structures.
• Automated analysis tools for SMLM image quantification and classification are currently lacking.
• Accurate classification of cellular structures is crucial for understanding biological processes.

Purpose of the Study:

• To introduce Enhanced Classification of Localized Point clouds by Shape Extraction (ECLiPSE), an automated machine learning pipeline for SMLM image analysis.
• To enable accurate classification of 2D and 3D cellular structures using SMLM data.
• To provide a robust tool for studying protein aggregates and mitochondrial morphology.

Main Methods:

• Developed ECLiPSE, a machine learning pipeline utilizing shape descriptors extracted directly from SMLM localizations.
• Employed both unsupervised and supervised classification methods for validation.
• Applied ECLiPSE to analyze protein aggregates in neurodegenerative disease models and to differentiate between healthy and depolarized mitochondria.

Main Results:

• ECLiPSE achieved near-perfect accuracy in classifying diverse cellular structures across validated datasets.
• The pipeline effectively classified morphologically distinct protein aggregates relevant to neurodegenerative diseases.
• 3D ECLiPSE successfully identified biological differences in mitochondrial morphology.

Conclusions:

• ECLiPSE offers a powerful, automated solution for classifying cellular structures from SMLM data.
• The tool enhances the study of nanoscale biological structures and their relevance in disease.
• ECLiPSE is expected to advance research across various biological contexts requiring high-resolution imaging analysis.