Merge pull request #11 from sina-mansour/dev

Dev

.github/workflows/release.yml

@@ -0,0 +1,28 @@
+name: Publish Python 🐍 distribution 📦 to PyPI upon new release
+
+on:
+  release:
+    types: [created]
+
+jobs:
+  build-n-publish:
+    name: Build and publish Python 🐍 distributions 📦 to PyPI
+    runs-on: ubuntu-latest
+
+    steps:
+    - uses: actions/checkout@v3
+    - name: Set up Python
+      uses: actions/setup-python@v4
+      with:
+        python-version: "3.10"
+    - name: Install dependencies
+      run: |
+        python -m pip install --upgrade pip
+        pip install setuptools wheel twine
+    - name: Build and publish
+      env:
+        TWINE_USERNAME: __token__
+        TWINE_PASSWORD: ${{ secrets.PYPI_API_TOKEN }}
+      run: |
+        python setup.py sdist bdist_wheel
+        twine upload dist/*

.gitignore

@@ -15,3 +15,6 @@ connectome-spatial-smoothing.sublime-workspace
 # test notebooks
 code/Connectome_Spatial_Smoothing/__pycache__/
 notebooks/temp.ipynb
+
+# ignore the pypi toke
+token.tmp

README.md

@@ -41,6 +41,16 @@ Then, you could simply use the package in your own code after importing:
 
 ---
 
+## Usage notes
+
+### Verbosity:
+
+By default CSS scripts print out a bunch of logs that may or may not be of interest to you. If you like to disable the logs, simply add the following script:
+
+`css._verbose = False`
+
+---
+
 We have provided a short jupyter notebook showcasing all the functionalities described above. You may use the following link to open [this notebook](https://github.com/sina-mansour/connectome-based-smoothing/blob/main/notebooks/example.ipynb) in an interactive google colab session:
 
 [![Open In Colab](https://colab.research.google.com/assets/colab-badge.svg)](https://colab.research.google.com/github/sina-mansour/connectome-based-smoothing/blob/main/notebooks/example.ipynb)

code/Connectome_Spatial_Smoothing/CSS.py

@@ -29,6 +29,7 @@
 _main_dir = os.path.abspath(os.path.dirname(__file__))
 _sample_cifti_dscalar = os.path.join(_main_dir, 'data/templates/cifti/ones.dscalar.nii')
 _glasser_cifti = os.path.join(_main_dir, 'data/templates/atlas/Glasser360.32k_fs_LR.dlabel.nii')
+_verbose=True
 
 
 def _join_path(*args):
@@ -53,12 +54,13 @@ def _time_str(mode='abs', base=None):
 
 
 def _print_log(message, mode='info'):
-    if mode == 'info':
-        print ('{}: \033[0;32m[INFO]\033[0m {}'.format(_time_str(), message))
-    if mode == 'err':
-        print ('{}: \033[0;31m[ERROR]\033[0m {}'.format(_time_str(), message))
-        quit()
-    sys.stdout.flush()
+    if _verbose:
+        if mode == 'info':
+            print ('{}: \033[0;32m[INFO]\033[0m {}'.format(_time_str(), message))
+        if mode == 'err':
+            print ('{}: \033[0;31m[ERROR]\033[0m {}'.format(_time_str(), message))
+            quit()
+        sys.stdout.flush()
 
 
 def _handle_process_with_que(que, func, args, kwds):
@@ -206,7 +208,7 @@ def parcellation_characteristic_matrix(atlas_file=_glasser_cifti, cifti_file=_sa
 
     label_dict = {x: i for (i, x) in enumerate(labels)}
 
-    parcellation_matrix = np.zeros((len(labels), cortical_vertices_count))
+    parcellation_matrix = np.zeros((len(labels), len(surface_labels)))
 
     for (i, x) in enumerate(surface_labels):
         parcellation_matrix[label_dict[x], i] = 1
@@ -601,11 +603,19 @@ def _get_half_incidence_matrices_from_endpoint_distances(start_dists,
                                                          end_dists,
                                                          end_indices,
                                                          node_count,
-                                                         threshold):
+                                                         threshold,
+                                                         weights=None):
     """
     Returns two half incidence matrices in a sparse format (CSR) after
     filtering the streamlines that are far (>2mm) from their closest vertex.
     """
+
+    if weights is None:
+        weights = np.ones(len(start_dists))
+    elif (type(weights) == str):
+        # load text file contents (this could be sift weights files generated by mrtrix)
+        weights = np.genfromtxt(weights)
+
     # mask points that are further than the threshold from all surface coordinates
     outlier_mask = (start_dists > threshold) | (end_dists > threshold)
     _print_log('outliers located: #{} outliers ({}%, with threshold {}mm)'.format(
@@ -620,7 +630,7 @@ def _get_half_incidence_matrices_from_endpoint_distances(start_dists,
     end_dict = {}
     indices = (i for i in range(len(outlier_mask)) if not outlier_mask[i])
     for l, i in enumerate(indices):
-        start_dict[(start_indices[i], l)] = start_dict.get((start_indices[i], l), 0) + 1
+        start_dict[(start_indices[i], l)] = start_dict.get((start_indices[i], l), 0) + weights[i]
         end_dict[(end_indices[i], l)] = end_dict.get((end_indices[i], l), 0) + 1
 
     start_inc_mat = sparse.dok_matrix(
@@ -650,11 +660,15 @@ def _get_half_incidence_matrices_from_endpoint_distances(start_dists,
     return (start_inc_mat.tocsr(), end_inc_mat.tocsr())
 
 
-def _get_adjacency_from_half_incidence_matrices(U, V):
+def _get_adjacency_from_half_incidence_matrices(U, V, stat='sum'):
     """
     return a sparse adjacency matrix A from the two halfs of incidence matrix U & V.
     """
     A = U.dot(V.T)
+    if stat == 'mean':
+        # compute average instead of sum
+        A_div = (U != 0).astype(int).dot(((V != 0).astype(int)).T)
+        A.data = A.data / A_div.data
     return A + A.T
 
 
@@ -664,7 +678,9 @@ def map_high_resolution_structural_connectivity(track_file,
                                                 warp_file=None,
                                                 threshold=2,
                                                 subcortex=False,
-                                                cifti_file=_sample_cifti_dscalar,):
+                                                cifti_file=_sample_cifti_dscalar,
+                                                weights=None,
+                                                stat='sum'):
     """
     Map the high-resolution structural connectivity matrix from tractography outputs.
 
@@ -689,6 +705,15 @@ def map_high_resolution_structural_connectivity(track_file,
                     to determine the high-resolution structure to exclude the medial wall and potentially
                     integrate subcortical volume.
 
+        weights: [optional] A numpy vector with the length of number of streamlines, or a file path
+                 pointing to a per-streamline weights file (such as the files generated by mrtrix's
+                 tcksift2). If this file is provided, then the weights are used to adjust the 
+                 contribution of every streamline to the connectivity matrix.
+        stat: [optional, default='sum'] Either 'sum', or 'mean'. If sum, the values will be added up
+              over all streamlines. Otherwise an average is computed over all streamline weights.
+              combination of stat='mean' with a list of per-streamline lengths could be used to compute
+              a mean distance connectivity matrix.
+
     Returns:
 
         connectome: The high-resolution structural connectome in a sparse csr format.
@@ -703,8 +728,10 @@ def map_high_resolution_structural_connectivity(track_file,
                 warp_file,
                 subcortex=subcortex,
             ),
-            threshold=threshold
-        )
+            threshold=threshold,
+            weights=weights
+        ),
+        stat=stat
     )
 
 

setup.py

@@ -7,7 +7,7 @@
 
 setuptools.setup(
     name="Connectome_Spatial_Smoothing",
-    version="0.1.3",
+    version="0.1.4",
     author="Sina Mansour L.",
     author_email="[email protected]",
     description="Connectome Spatial Smoothing",