Hidden Markov Modeling for Maximum Likelihood Neuron Reconstruction


Athey, T.L., Tward, D.J., Mueller, U. et al. Hidden Markov modeling for maximum probability neuron reconstruction. Commun Biol 5, 388 (2022). https://doi.org/10.1038/s42003-022-03320-0.

Relevant directory


How to use ViterBrain

  • First, make sure that you have installed the brainlit package [Documentation].

  • Second, uncompress the data brainlit/experiments/ViterBrain/data/example.zip. brainlit/experiemnts/ViterBrain/data/sample.zip can also be used.

  • Make sure you are using Python3.9

  • Then, you can run some of the tutorial notebooks in the notebooks folder:
    • ViterBrain.ipynb - shows a programmatic example of the pipeline, based on zarr inputs.

    • fig3-voxels.ipynb - generates Figure 3 from the paper.

    • fig7-results.ipynb - generates Figure 7 from the paper.

    • other notebooks can be useful for referemce, they were used in generating results in the paper.

  • The files in the scripts folder also can be useful:
    • napari_gui.py - shows the GUI prototype.
      • click on colored fragment to select, red arrow will identify orientation.

      • o-key or switch states button to switch orientation of selected fragment.

      • click on another colored fragment (and hit o-key if necessary to switch orientation).

      • click on the labels layer in the left hand pane, then click somewhere on the image (not on a fragment)

      • t-key or trace button to trace between fragments.

      • c-key or clear selected states button to clear the selected fragments.

      • q-key or clear all button to clear all annotations.

      • n-key or next color button to change colors (3 total colors).

    • other scripts are for reference for benchmarking the timing of the pipeline.