Merge pull request #17 from samuelstjean/maint

Maint

.github/workflows/python-package.yml

@@ -15,20 +15,19 @@ jobs:
     runs-on: ubuntu-latest
     strategy:
       matrix:
-        python-version: ["3.6", "3.7", "3.8", "3.9", "3.10"]
+        python-version: ["3.7", "3.8", "3.9", "3.10", "3.11"]
         os: [ubuntu-latest, macos-latest, windows-latest]
 
     steps:
-    - uses: actions/checkout@v2
+    - uses: actions/checkout@v3
     - name: Set up Python ${{ matrix.python-version }}
-      uses: actions/setup-python@v2
+      uses: actions/setup-python@v4
       with:
         python-version: ${{ matrix.python-version }}
     - name: Install dependencies
       run: |
         python -m pip install --upgrade pip
         pip install flake8 pytest
-        if [ -f requirements.txt ]; then pip install -r requirements.txt; fi
         pip install .
     - name: Lint with flake8
       run: |

.github/workflows/pythonpublish.yml

@@ -8,9 +8,9 @@ jobs:
   deploy:
     runs-on: ubuntu-latest
     steps:
-    - uses: actions/checkout@v1
+    - uses: actions/checkout@v3
     - name: Set up Python
-      uses: actions/setup-python@v1
+      uses: actions/setup-python@v4
       with:
         python-version: '3.x'
     - name: Install dependencies

CHANGELOG.md

@@ -1,9 +1,15 @@
 # Changelog
 
+## [v0.2.6] - 2022-11-04
+
+- Change the main script name from **get_distribution** to **autodmri_get_distribution** for consistency with the package name.
+- Remove python 2.7 support.
+- Fix some deprecation warnings from numpy.
+
 ## [v0.2.5] - 2021-03-09
 
 - Remove nans from the output when using noise maps.
-- Fix the main script regarding the noise maps to 4d conversion.
+- Fix the main script regarding the noise maps to 4D conversion.
 - Remove Dockerfile.
 
 ## [v0.2.4] - 2021-03-09

README.md

@@ -13,6 +13,14 @@ pip install autodmri
 
 You can find a quick example and datasets over [here](example) and the full documentation at http://autodmri.rtfd.io/.
 
+An example, super basic call to the script would be
+
+~~~bash
+autodmri_get_distribution dwi.nii.gz sigma.nii.gz N.nii.gz mask.nii.gz
+~~~
+
+Be sure to check the options by passing `--help` to the script.
+
 ###  The manuscript and references
 
 You can read the [journal version][media] in Medical Image Analysis and the datasets are available here https://zenodo.org/record/2483105.

autodmri/blocks.py

@@ -1,5 +1,3 @@
-from __future__ import division
-
 import numpy as np
 import numbers
 

autodmri/estimator.py

@@ -1,5 +1,3 @@
-from __future__ import division
-
 import numpy as np
 
 from scipy.ndimage.interpolation import zoom
@@ -112,8 +110,8 @@ def get_mask(data, N_min, N_max, phi, alpha_prob=0.05):
         lambda_minus = lambda_cdf(N_min*K, alpha_prob/2)
         lambda_plus = lambda_cdf(N_max*K, 1 - alpha_prob/2)
 
-        mask_current = np.zeros(data.shape[:-1], dtype=np.bool)
-        mask_loop = np.zeros(data.shape[:-1], dtype=np.bool)
+        mask_current = np.zeros(data.shape[:-1], dtype=bool)
+        mask_loop = np.zeros(data.shape[:-1], dtype=bool)
 
         for sigma in phi:
             s = sum_data2 / (2*sigma**2)
@@ -126,7 +124,7 @@ def get_mask(data, N_min, N_max, phi, alpha_prob=0.05):
 
     # Explicitly remove known artifacts
     if exclude_mask is None:
-        exclude_mask = np.zeros(data.shape[:-1], dtype=np.bool)
+        exclude_mask = np.zeros(data.shape[:-1], dtype=bool)
 
     # we don't know N, so guess parameters iteratively
     data = data.astype(np.float64)  # prevent data**4 overflow
@@ -201,7 +199,7 @@ def estimate_from_nmaps(data, size=5, return_mask=True, method='moments', full=F
     -------
     sigma, N, mask (optional)
     '''
-    m_out = np.zeros(data.shape[:-1], dtype=np.bool)
+    m_out = np.zeros(data.shape[:-1], dtype=bool)
     median = np.median(data)
 
     if median == 0:
@@ -242,7 +240,7 @@ def estimate_from_nmaps(data, size=5, return_mask=True, method='moments', full=F
 
         sigma = np.zeros(reshaped_maps.shape[0], dtype=np.float32)
         N = np.zeros(reshaped_maps.shape[0], dtype=np.float32)
-        mask = np.zeros((reshaped_maps.shape[0], size**3), dtype=np.bool)
+        mask = np.zeros((reshaped_maps.shape[0], size**3), dtype=bool)
 
         output = Parallel(n_jobs=ncores,
                           verbose=verbose)(delayed(proc_inner)(reshaped_maps[i], median, size, method, use_rejection) for i in range(reshaped_maps.shape[0]))
@@ -280,7 +278,7 @@ def proc_inner(cur_map, median, size, method, use_rejection):
     else:
         cur_map = cur_map.reshape(size**3, 1, -1)
         sigma, N = get_noise_distribution(cur_map, method=method)
-        mask = np.ones_like(cur_map, dtype=np.bool)
+        mask = np.ones_like(cur_map, dtype=bool)
 
         if np.isnan(sigma) or np.isnan(N):
             sigma = 0

autodmri/gamma.py

@@ -1,5 +1,3 @@
-from __future__ import division
-
 import numpy as np
 
 from scipy.special import digamma, polygamma
@@ -39,7 +37,7 @@ def get_noise_distribution(data, method='moments'):
     elif method == 'maxlk':
         sigma = maxlk_sigma(data)
     else:
-        raise ValueError('Invalid method name {}'.format(method))
+        raise ValueError(f'Invalid method name {method}')
 
     t = data**2 / (2*sigma**2)
 
@@ -50,7 +48,7 @@ def get_noise_distribution(data, method='moments'):
         y = np.mean(np.log(t))
         N = inv_digamma(y)
     else:
-        raise ValueError('Invalid method name {}'.format(method))
+        raise ValueError(f'Invalid method name {method}')
 
     return sigma, N
 

example/README.md

@@ -6,7 +6,7 @@ You can change things such as the number of cores for parallel processing, which
 the size of the window, to do a full estimation or only non overlapping windows for noise maps, etc.
 
 ~~~bash
-get_distribution -h
+autodmri_get_distribution -h
 ~~~
 
 Further information is available from the [documentation](https://autodmri.readthedocs.io/en/stable/) as these are more advanced cases.
@@ -31,7 +31,7 @@ The data looks like this
 1. With equations using the moments (the default, also accessible with `-m moments`)
 
 ~~~bash
-get_distribution dwi_1_8.nii.gz sigma.nii.gz N.nii.gz mask.nii.gz
+autodmri_get_distribution dwi_1_8.nii.gz sigma.nii.gz N.nii.gz mask.nii.gz
 ~~~
 
 And the output files looks like that.
@@ -48,7 +48,7 @@ so if we look at an axial view we would only see a single value while the other
 2. With equations using the maximum likelihood (with the option `-m maxlk`)
 
 ~~~bash
-get_distribution dwi_1_8.nii.gz sigma_maxlk.nii.gz N_maxlk.nii.gz mask_maxlk.nii.gz -m maxlk
+autodmri_get_distribution dwi_1_8.nii.gz sigma_maxlk.nii.gz N_maxlk.nii.gz mask_maxlk.nii.gz -m maxlk
 ~~~
 
 ![](sigma_maxlk.png)
@@ -69,7 +69,7 @@ Notice how every output is now a full fledged 3D maps.
 1. With equations using the moments
 
 ~~~bash
-get_distribution data_SENSE3_MB3_noisemap.nii.gz sigma_nmaps.nii.gz N_nmaps.nii.gz mask_nmaps.nii.gz --noise_maps
+autodmri_get_distribution data_SENSE3_MB3_noisemap.nii.gz sigma_nmaps.nii.gz N_nmaps.nii.gz mask_nmaps.nii.gz --noise_maps
 ~~~
 
 ![](sigma_nmaps.png)
@@ -80,7 +80,7 @@ get_distribution data_SENSE3_MB3_noisemap.nii.gz sigma_nmaps.nii.gz N_nmaps.nii.
 2. With equations using the maximum likelihood
 
 ~~~bash
-get_distribution data_SENSE3_MB3_noisemap.nii.gz sigma_maxlk_nmaps.nii.gz N_maxlk_nmaps.nii.gz mask_maxlk_nmaps.nii.gz -m maxlk --noise_maps
+autodmri_get_distribution data_SENSE3_MB3_noisemap.nii.gz sigma_maxlk_nmaps.nii.gz N_maxlk_nmaps.nii.gz mask_maxlk_nmaps.nii.gz -m maxlk --noise_maps
 ~~~
 
 ![](sigma_maxlk_nmaps.png)

pyproject.toml

@@ -1,3 +1,3 @@
 [build-system]
-requires = ["setuptools"]
+requires = ["setuptools>=61.0"]
 build-backend = "setuptools.build_meta"

scripts/get_distribution → scripts/autodmri_get_distribution

@@ -1,7 +1,5 @@
 #!/usr/bin/env python
 
-from __future__ import division
-
 import numpy as np
 import nibabel as nib
 
@@ -122,9 +120,9 @@ def main():
     for f in overwritable_files:
         if f is not None and os.path.isfile(f):
             if args.overwrite:
-                logger.warning('Overwriting {}'.format(os.path.realpath(f)))
+                logger.warning(f'Overwriting {os.path.realpath(f)}')
             else:
-                parser.error('{} already exists! Use -f or --force to overwrite it.'.format(f))
+                parser.error(f'{f} already exists! Use -f or --force to overwrite it.')
 
     vol = nib.load(args.data)
     data = vol.get_fdata(dtype=np.float32)
@@ -139,23 +137,23 @@ def main():
     noise_maps = args.noise_maps
 
     if args.exclude is not None:
-        exclude_mask = nib.load(args.exclude).get_fdata().astype(np.bool)
-        logger.info('Excluding voxels from file {}'.format(args.exclude))
+        exclude_mask = nib.load(args.exclude).get_fdata().astype(bool)
+        logger.info(f'Excluding voxels from file {args.exclude}')
     else:
         exclude_mask = None
 
-    logger.info('Now estimating over file {} with method = {} and axis = {}'.format(args.data, method, axis))
+    logger.info(f'Now estimating over file {args.data} with method = {method} and axis = {axis}')
 
     if noise_maps:
         if full:
             overlap = 'overlapping windows'
         else:
-            overlap = 'non overlapping windows'
+            overlap = 'non-overlapping windows'
 
         if data.ndim == 3:
             data = data[..., None]
 
-        logger.info('Estimation will be done over noise maps with a window of size {} and {}'.format(size, overlap))
+        logger.info(f'Estimation will be done over noise maps with a window of size {size} and {overlap}')
         sigma, N, mask = estimate_from_nmaps(data, size=size, return_mask=True, method=method, full=full, ncores=ncores, use_rejection=False, verbose=args.verbose)
 
     else:
@@ -165,7 +163,7 @@ def main():
         if args.fast_median:
             logger.info('Estimation of the medians will be done over each volume, then on the median of the medians.')
         elif data.shape[-1] > 100:
-            logger.warning('Estimation of the median will be done over the whole volume, but you have {} volumes.\n'.format(data.shape[-1]) +
+            logger.warning(f'Estimation of the median will be done over the whole volume, but you have {data.shape[-1]} volumes.\n' +
                            '\tConsider the option --fast_median if memory usage is high and startup time is too long.')
 
         sigma, N, mask = estimate_from_dwis(data, axis=axis, return_mask=True, exclude_mask=exclude_mask, ncores=ncores,
@@ -182,13 +180,13 @@ def main():
             sigma = sigma[None, None, :]
             N = N[None, None, :]
         else:
-            raise ValueError('axis = {} is not 0, 1 or 2, but that should never happen!'.format(axis))
+            raise ValueError(f'axis = {axis} is not 0, 1 or 2, but that should never happen!')
 
         sigma = np.ones(mask.shape) * sigma
         N = np.ones(mask.shape) * N
 
     # Save the data
-    logger.info('Output files are {}, {} and {}'.format(args.sigma, args.N, args.mask))
+    logger.info(f'Output files are {args.sigma}, {args.N} and {args.mask}')
     mask = mask.astype(np.int16)
 
     nib.Nifti1Image(sigma, aff).to_filename(args.sigma)

setup.py

@@ -1,15 +1,14 @@
-import io
 from setuptools import setup, find_packages
 
-with io.open('README.md', encoding='utf-8') as f:
+with open('README.md', encoding='utf-8') as f:
     long_description = f.read()
 
 setup(
     name='autodmri',
-    version='0.2.5',
+    version='0.2.6',
     author='Samuel St-Jean',
     packages=find_packages(),
-    scripts=['scripts/get_distribution'],
+    scripts=['scripts/autodmri_get_distribution'],
     url='https://github.com/samuelstjean/autodmri',
     license='MIT',
     description='Implementation of "Automated characterization of noise distributions in diffusion MRI data".',