I am trying to use the data_per_streamline attribute to select streamlines from within a Tractogram object. I am running into the following behavior.

I initialize a very simple tractogram object. The first streamline has the data_per_streamline foo key set to 0 and the other two streamlines have this set to 1:

import numpy as np import nibabel.streamlines as sl tgram = sl.Tractogram([np.array([[0, 0, 0], [1, 1, 1]]), np.array([[2, 2, 2], [3, 3, 3], [4, 4, 4]]), np.array([[5, 5, 5], [6, 6, 6]])], data_per_streamline={'foo': [[0],[1],[1]]}) 

I can find the right streamlines using np.where:

idx = np.where(tgram.data_per_streamline['foo'] == 1)[0] idx array([1, 2]) 

It even looks like it's doing the right thing when I use this for indexing:

tgram[idx].streamlines ArraySequence([array([[2, 2, 2], [3, 3, 3], [4, 4, 4]]), array([[5, 5, 5], [6, 6, 6]])]) 

But under the hood:

tgram[idx].streamlines._data array([[0, 0, 0], [1, 1, 1], [2, 2, 2], [3, 3, 3], [4, 4, 4], [5, 5, 5], [6, 6, 6]]) 

Why is it still holding onto all of the data? Is there some way to drop the data for the first (not selected) streamline?