Zebrafish Neuronal Analysis by Bioimage Informatics
Division of molecular and developmental biology of NIG is studying the activity in the brain and neuronal system of zebrafish. They use calcium imaging techniques to quantify the temporal and spatial activity of neurons from time series images, and model the behavior of the brain using machine learning. Required techniques are 2D registration, gray-scale segmentation, time-series analysis, object tracking, and so on.
The project is challenging the following problems.
- Image registration - a live zebrafish makes motion during observations, so this motion should be removed by registration.
- Neuron detection - the location of neurons appear in images should be comprehensively detected regardless of their activity.
- Collection of neuronal activity signal - the activity of neruons are collected from detected ones to make time-series data.
- Characterization of signal - which neuron is active for which period is characterized quantitatively using time-series analysis.
- Relationship between neurons - which neurons are related is quantitatively estimated using spatio-temporal analysis.
Some of our research results are released as tools.