A tool for visualizing and segmenting and identifying neural activity from 3d volumetric optical recordings. Specifically designed for neuronal responses to odor.
- Estimates forground voxels and performs nonlinear embedding (t-SNE) on voxels to visualize the temporal structure (can also use spatial stucture as well).
Forground is to the left, 2D embedding on the right. - Provides manual and automatic clustering options on both the 2D embedded space (right) and on the physical space (left).
- Can compare spatial footprint and activities of different neurons for validation and visualization.
- In cases where training data exist, can classify neurons to particular neuronal IDs using SVM based on odor response. Results of analysis/manual proof-reading can contribute to training dataset.
- Generates 3D reconstruction of neuronal footprints
- ELKI (download here: http://elki.dbs.ifi.lmu.de/) - main ELKI folder must be on matlab path
- My version of Barnes-Hut t-SNE - https://github.com/jacobbaron/bhtsne - follow instructions to build .exe file. The .exe file must be in the same folder as the fast-tsne matlab function, and all must be on matlab path.
- CIA (https://github.com/jacobbaron/CIA): Calcium Imaging Analysis - Imports ND2 or HDF5 files, HDF5 files are generated by Andor Acquire (https://github.com/jacobbaron/AndorAcquire)
- NoRMCorre (https://github.com/flatironinstitute/NoRMCorre) - Package for rigid/nonrigid alignment of neuronal recordings.
- For SVM classifier, Python 2.7, SKlearn.