Data downloads and browsers

Guide to using the AnnoJ browser

Software downloads and documentation

CellSort: Automated cell sorting for dynamic fluorescence microscopy

Cell Sorting
Optical fluorescence microscopy allows neuroscientists to observe and record electrical and chemical activity of brain cells in behaving animals. Techniques such as calcium-dependent fluorscence imaging generate movie data sets, in which information about neural and glial activity is distributed across many pixels and time frames. CellSort is a computational procedure designed to efficiently identify cellular signals by finding combinations of pixels with statistically independent dynamics. The algorithm combines principal and independent component analysis (PCA, ICA) as well as image segmentation and temporal deconvolution. There is a nice overview of the challenge and discussion of our approach by Alex Kwan in the HFSP Journal.

The image at left illustrates the results of our cell sorting procedure, which estimates cellular spatial filters (top left), calcium-dependent fluorescence dynamics (bottom left), as well as the underlying spike trains (bottom right).

The movie (below) shows fluorescence imaging of calcium dynamics in dendrites of mouse cerebellar Purkinje neurons, recorded by in vivo two-photon microscopy. Left panel: Raw data, normalized by the average fluorescence in each pixel (ΔF/F). Right panel: Action potentials within each cell are detected by CellSort. The soundtrack was generated by assigning each dendrite to a different piano key, arranged from the low (upper left) to the high notes (lower right).

DeconSTORM: Analysis of Super-Resolution Fluorescence Microscopy by Statistical Deconvolution

Cell Sorting
DeconSTORM is a tool for analysis of super-resolution microscopy data sets from of stochastic optical reconstruction micrscopy (STORM), photoactivated localization microscopy (PALM), and related techniques. Our computational analysis approach is described in this publication, and a MATLAB software toolbox is available for implementing our algorithm.