Blind demixing methods for recovering dense neuronal morphology from barcode imaging data
Shuonan Chen1,2,3,4,5,6, Jackson Loper1,2,3,4,5,7, Pengcheng Zhou8,9
1Mortimer B. Zuckerman Mind Brain Behavior Institute, Columbia University, New York, New York, United States of America.
Plos Computational Biology
|April 8, 2022
Summary
Cellular barcoding enables tracing individual neurons by assigning unique codes. Accurate reconstruction is feasible with high signal density, allowing large-scale neuronal mapping using light microscopy.
Area of Science:
- Neuroscience
- Computational Biology
- Microscopy
Background:
- Cellular barcoding offers potential for tracing individual neuronal microanatomy.
- Current methods face challenges in reconstructing complex neuronal structures.
Purpose of the Study:
- To quantify the feasibility of 'infinite-pseudocolor' cellular barcoding for neuronal reconstruction.
- To develop and validate computational methods for barcode recovery and neuronal morphology reconstruction.
Main Methods:
- Simulations based on electron microscopy data with signal structure matched to real barcoding data.
- Development of a blind demixing algorithm for barcode recovery.
- Application of a neural network for reconstructing neuronal morphology from barcode signals.
Main Results:
- Accurate recovery of neuronal barcodes is feasible with sufficiently high signal density.
- The developed blind demixing method was validated on real data.
- A neural network successfully connected discontiguous barcode signals to reconstruct morphology.
Conclusions:
- Cellular barcoding is a promising technique for high-resolution, large-scale neuronal mapping.
- The developed computational tools enhance the feasibility of this approach.
- This method could revolutionize the study of neuronal morphology and projection patterns.
