fafbseg.google.autoreview_edges#

fafbseg.google.autoreview_edges(x, conf_threshold=1, vol=None, remote_instance=None)[source]#

Automatically review (low-confidence) edges between nodes.

The way this works:

  1. Fetch the live version of the neuron(s) from the CATMAID instance

  2. Use raycasting to test (low-confidence) edges

  3. Edge confidence is set to 5 if test is passed and to 1 if not

You can use this function to test all edges in a neuron by increasing conf_threshold to 5. Please note that this could produce a lot of false positives (i.e. edges will be flagged as incorrect even though they aren’t). Part of the problem is that mitochondria are segmented as separate entities and hence introduce membranes inside a neuron.

Parameters:

  • x (skeleton ID(s) | pymaid.CatmaidNeuron/List) – Neuron(s) to review.

  • conf_threshold (int, optional) – Confidence threshold for edges to be tested. By default only reviews edges with confidence <= 1.

  • vol (cloudvolume.CloudVolume, optional) – CloudVolume pointing to segmentation data.

  • remote_instance (pymaid.CatmaidInstance, optional) – CATMAID instance. If None, will use globally define instance.

Returns:

CATMAID server response from updating node confidences.

Return type:

server response

See also

fafbseg.test_edges()

If you only need to test without changing confidences.

Examples

>>> # Set up CloudVolume from the publicly hosted FAFB segmentation data
>>> # (if you have a local copy, use that instead)
>>> from cloudvolume import CloudVolume
>>> vol = CloudVolume('https://storage.googleapis.com/fafb-ffn1-20190805/segmentation',
...                   cache=True,
...                   progress=False)
>>> # Autoreview edges
>>> _ = fafbseg.autoreview_edges(14401884, vol=vol, remote_instance=manual)