Adding model to segment bones

.flake8

@@ -2,7 +2,7 @@
 # Keep in sync with setup.cfg and pyproject.toml which is used for source packages.
 
 [flake8]
-ignore = W503, E203, B950, B011, B904
+ignore = W503, E203, B950, B011, B904, B907, B905
 max-line-length = 100
 max-complexity = 18
 select = B,C,E,F,W,T4,B9

.github/workflows/ci.yml

@@ -15,7 +15,7 @@ jobs:
     runs-on: ubuntu-latest
     strategy:
       matrix:
-        python-version: ["3.6", "3.7", "3.8"]
+        python-version: ["3.7", "3.8", "3.9", "3.10"]
 
     steps:
       - uses: actions/checkout@v2
@@ -43,7 +43,7 @@ jobs:
     runs-on: ubuntu-latest
     strategy:
       matrix:
-        python-version: ["3.6", "3.7", "3.8"]
+        python-version: ["3.8", "3.9", "3.10"]
 
     steps:
       - uses: actions/checkout@v2
@@ -72,7 +72,7 @@ jobs:
     strategy:
       matrix:
         os: [ubuntu-latest]
-        python-version: ["3.6", "3.7", "3.8"]  # there are some issues with numpy multiarray in 3.7
+        python-version: ["3.7", "3.8", "3.9", "3.10"]  # there are some issues with numpy multiarray in 3.7
 
     steps:
       - uses: actions/checkout@v2
@@ -83,15 +83,13 @@ jobs:
           python-version: ${{ matrix.python-version }}
 
       - name: Install Dependencies
-        # there are some issues with numpy multiarray in 3.7, affecting numba 0.54 installation
-        # SimpleITK 2.1.0 does not support non-orthonormal directions
         run: |
           python -m pip install --upgrade pip
-          pip install numba==0.53.1
-          pip install tensorflow==2.4.1 keras
+          pip install numba
+          pip install tensorflow>=2.0.0
           pip install torch
           pip install sigpy
-          pip install --upgrade simpleitk==2.0.2
+          pip install --upgrade simpleitk
           make dev
           pip install -e '.[dev]'
 
@@ -112,6 +110,7 @@ jobs:
           make test-cov
 
       - name: Upload to codecov.io
+        if: github.event_name != 'pull_request' || github.event.pull_request.head.repo.full_name == github.repository
         uses: codecov/codecov-action@v1
         with:
           file: ./coverage.xml

.gitignore

@@ -54,4 +54,13 @@ docs/source/generated
 
 # coverage files
 .coverage
-coverage.xml
+coverage.xml
+
+# system files
+.DS_Store
+
+#nipype logfiles
+stdout.nipype
+
+# ignore "scratch" files used for testing / debugging
+**/scratch

Makefile

@@ -30,6 +30,7 @@ build-docs:
 	cd docs && make html
 
 dev:
+	pip install pytest
 	pip install black==21.4b2 click==8.0.2 coverage isort flake8 flake8-bugbear flake8-comprehensions
 	pip install --upgrade mistune==0.8.4 sphinx sphinx-rtd-theme recommonmark m2r2
 	pip install -r docs/requirements.txt

README.md

@@ -14,7 +14,7 @@ DOSMA is an AI-powered Python library for medical image analysis. This includes,
 We hope that this open-source pipeline will be useful for quick anatomy/pathology analysis and will serve as a hub for adding support for analyzing different anatomies and scan sequences.
 
 ## Installation
-DOSMA requires Python 3.6+. The core module depends on numpy, nibabel, nipype,
+DOSMA requires Python 3.7+. The core module depends on numpy, nibabel, nipype,
 pandas, pydicom, scikit-image, scipy, PyYAML, and tqdm.
 
 Additional AI features can be unlocked by installing tensorflow and keras. To
@@ -30,6 +30,24 @@ pip install dosma
 pip install dosma[ai]
 ```
 
+If you would like to install from source - to get the latest versions, or to build for your specific system: 
+
+If you would like to install from source then use: 
+
+```bash
+git clone git@github.com:ad12/DOSMA.git
+cd DOSMA
+pip install .
+```
+
+To get AI support: 
+
+```bash
+git clone git@github.com:ad12/DOSMA.git
+cd DOSMA
+pip install '.[ai]'
+```
+
 If you would like to contribute to DOSMA, we recommend you clone the repository and
 install DOSMA with `pip` in editable mode.
 
@@ -41,6 +59,10 @@ make dev
 ```
 
 To run tests, build documentation and contribute, run
+- If Elastix not installed, must indicate this to system by running the following on the commandline: 
+   `export "DOSMA_UNITTEST_DISABLE_ELASTIX=True"`
+- If test data not available add:
+   `export ""DOSMA_UNITTEST_DISABLE_DATA"=True"`
 ```bash
 make autoformat test build-docs
 ```

docs/requirements.txt

@@ -1,4 +1,4 @@
-h5py<3.0.0
+h5py
 numpy
 natsort
 nested-lookup

dosma/cli.py

@@ -33,8 +33,8 @@
 from dosma.core.io.format_io import ImageDataFormat
 from dosma.core.quant_vals import QuantitativeValueType as QV
 from dosma.defaults import preferences
+from dosma.models.model_loading_util import SUPPORTED_MODELS, get_model, model_from_config
 from dosma.models.seg_model import SegModel
-from dosma.models.util import SUPPORTED_MODELS, get_model, model_from_config
 from dosma.msk import knee
 from dosma.scan_sequences.mri.cones import Cones
 from dosma.scan_sequences.mri.cube_quant import CubeQuant
@@ -134,7 +134,8 @@ def __init__(
         if (dicom_path is not None) and (not os.path.isdir(dicom_path)):
             if load_path is not None:
                 warnings.warn(
-                    "Dicom_path {} not found. Will load data from {}".format(dicom_path, load_path)
+                    "Dicom_path {} not found. Will load data from {}".format(dicom_path, load_path),
+                    stacklevel=2,
                 )
             else:
                 raise NotADirectoryError("{} is not a directory".format(dicom_path))

dosma/core/fitting.py

@@ -69,7 +69,8 @@ def _format_out_ufuncs(self, _out_ufuncs, _func_nparams):
             if len(_out_ufuncs) > _func_nparams:
                 warnings.warn(
                     f"len(out_ufuncs)={len(_out_ufuncs)}, but only {_func_nparams} parameters. "
-                    f"Extra ufuncs will be ignored."
+                    f"Extra ufuncs will be ignored.",
+                    stacklevel=2,
                 )
 
         return _out_ufuncs
@@ -174,7 +175,7 @@ def fit(self, x, y: Sequence[MedicalVolume], mask=None, copy_headers: bool = Tru
         svs = []
 
         if (not isinstance(y, (list, tuple))) or (
-            not all([isinstance(_y, MedicalVolume) for _y in y])
+            not all(isinstance(_y, MedicalVolume) for _y in y)
         ):
             raise TypeError("`y` must be sequence of MedicalVolumes.")
 
@@ -647,15 +648,17 @@ def __init__(
         if y is not None:
             warnings.warn(
                 f"Setting `y` in the constructor can result in significant memory overhead. "
-                f"Specify `y` in `{type(self).__name__}.fit(y=...)` instead."
+                f"Specify `y` in `{type(self).__name__}.fit(y=...)` instead.",
+                stacklevel=2,
             )
             self._check_y(x, y)
         self.y = y
 
         if mask is not None:
             warnings.warn(
                 f"Setting `mask` in the constructor can result in significant memory overhead. "
-                f"Specify `mask` in `{type(self).__name__}.fit(mask=...)` instead."
+                f"Specify `mask` in `{type(self).__name__}.fit(mask=...)` instead.",
+                stacklevel=2,
             )
         self.mask = mask
 
@@ -739,7 +742,7 @@ def fit(self, x=None, y: Sequence[MedicalVolume] = None, mask=None):
         return tc_map, r_squared
 
     def _check_y(self, x, y):
-        if (not isinstance(y, Sequence)) or (not all([isinstance(sv, MedicalVolume) for sv in y])):
+        if (not isinstance(y, Sequence)) or (not all(isinstance(sv, MedicalVolume) for sv in y)):
             raise TypeError("`y` must be list of MedicalVolumes.")
 
         if any(x.device != cpu_device for x in y):
@@ -844,7 +847,9 @@ def curve_fit(
 
     oob = y_bounds is not None and ((y < y_bounds[0]).any() or (y > y_bounds[1]).any())
     if oob:
-        warnings.warn("Out of bounds values found. Failure in fit will result in np.nan")
+        warnings.warn(
+            "Out of bounds values found. Failure in fit will result in np.nan", stacklevel=2
+        )
 
     y_T = y.T
     if p0_seq:
@@ -963,7 +968,9 @@ def _compute_r2_matrix(_x, _y, _popts):
 
     oob = y_bounds is not None and ((y < y_bounds[0]).any() or (y > y_bounds[1]).any())
     if oob:
-        warnings.warn("Out of bounds values found. Failure in fit will result in np.nan")
+        warnings.warn(
+            "Out of bounds values found. Failure in fit will result in np.nan", stacklevel=2
+        )
 
     fitter = partial(
         _polyfit, x=x, deg=deg, y_bounds=y_bounds, rcond=rcond, w=w, eps=eps, xp=xp.__name__
@@ -1137,7 +1144,7 @@ def _format_p0(p0, param_args, N):
                 f"`p0` has unknown keys: {extra_keys}. "
                 f"Function signature has parameters {param_args}."
             )
-        p0_default = {p: 1.0 for p in param_args}
+        p0_default = {p: 1.0 for p in param_args}  # noqa C420
         # Update p0_default to keep the parameter keys in order.
         p0_default.update(p0)
         p0 = p0_default
@@ -1170,6 +1177,7 @@ def func(t, a, b):
         "__fit_mono_exp__ is deprecated since v0.0.12 and will no longer be "
         "supported in v0.0.13. Use `curve_fit` instead.",
         DeprecationWarning,
+        stacklevel=2,
     )
 
     x = np.asarray(x)
@@ -1191,6 +1199,7 @@ def __fit_monoexp_tc__(x, ys, tc0, show_pbar=False):  # pragma: no cover
         "__fit_monoexp_tc__ is deprecated since v0.12 and will no longer be "
         "supported in v0.13. Use `curve_fit` instead.",
         DeprecationWarning,
+        stacklevel=2,
     )
 
     p0 = (1.0, -1 / tc0)
@@ -1203,7 +1212,8 @@ def __fit_monoexp_tc__(x, ys, tc0, show_pbar=False):  # pragma: no cover
         if (y < 0).any() and not warned_negative:
             warned_negative = True
             warnings.warn(
-                "Negative values found. Failure in monoexponential fit will result in np.nan"
+                "Negative values found. Failure in monoexponential fit will result in np.nan",
+                stacklevel=2,
             )
 
         # Skip any negative values or all values that are 0s

dosma/core/io/dicom_io.py

@@ -262,15 +262,15 @@ def load(
             )
 
         if self.num_workers:
-            fn = functools.partial(pydicom.read_file, force=True)
+            fn = functools.partial(pydicom.dcmread, force=True)
             if self.verbose:
                 dicom_slices = process_map(fn, lstFilesDCM, max_workers=self.num_workers)
             else:
                 with mp.Pool(self.num_workers) as p:
                     dicom_slices = p.map(fn, lstFilesDCM)
         else:
             dicom_slices = [
-                pydicom.read_file(fp, force=True)
+                pydicom.dcmread(fp, force=True)
                 for fp in tqdm(lstFilesDCM, disable=not self.verbose)
             ]