Address linting and test errors

.github/workflows/test-ci.yml

@@ -11,8 +11,8 @@ on:
 jobs:
   linter-check:
     runs-on: ubuntu-latest
-    steps: 
-      - name: Checkout 
+    steps:
+      - name: Checkout
         uses: actions/checkout@v3
       - name: Set up Python "3.10"
         uses: actions/setup-python@v4
@@ -26,8 +26,8 @@ jobs:
         run: black . --diff --color
       - name: ruff Check
         shell: bash
-        run: ruff src
-        
+        run: ruff check src
+
   test-suite:
     runs-on: ${{ matrix.os }}
     needs: linter-check
@@ -36,7 +36,7 @@ jobs:
       matrix:
         os: [ubuntu-latest]
         python-version: ["3.10", "3.8"]
-        
+
     steps:
       - uses: actions/checkout@v3
       - name: Set up Python ${{ matrix.python-version }}
@@ -51,5 +51,5 @@ jobs:
           python -m pip install .[test,optional]
       - name: Run Tests
         shell: bash
-        run: | 
+        run: |
           pytest -n auto -x

docs/denoisers.rst

@@ -1,10 +1,10 @@
-LLR Denosing methods
+LLR Denoising Methods
 =====================
 
 Patch-denoise implemement several local-low-rank methods, based on singular values thresholding.
 
 
-Singular Value thresholding
+Singular Value Thresholding
 ---------------------------
 
 General Procedure

pyproject.toml

@@ -52,19 +52,19 @@ omit = ["*tests*"]
 precision = 2
 exclude_lines = ["pragma: no cover", "raise NotImplementedError"]
 
-
 # Formatting using black.
 [tool.black]
 
 #linting using ruff
 [tool.ruff]
 src = ["src", "tests"]
-select = ["E", "F", "B", "Q", "UP", "D"]
 
+[tool.ruff.lint]
 ignore = ["B905"]
 exclude = ["examples/", "tests/"]
+select = ["E", "F", "B", "Q", "UP", "D"]
 
-[tool.ruff.pydocstyle]
+[tool.ruff.lint.pydocstyle]
 convention="numpy"
 
 [tool.isort]

src/patch_denoise/bindings/cli.py

@@ -177,7 +177,8 @@ def main():
     ]:
         extra_kwargs["noise_std"] = noise_map
         if noise_map is None:
-            raise RuntimeError("A noise map must me specified for this method.")
+            raise RuntimeError("A noise map must be specified for this method.")
+
     denoised_data, patchs_weight, noise_std_map, rank_map = denoise_func(
         input_data,
         patch_shape=d_par.patch_shape,

src/patch_denoise/bindings/modopt.py

@@ -49,7 +49,7 @@ def __init__(
             **kwargs,
         )
         self.op = self._op_method
-        self.cost = lambda *args, **kw: np.NaN
+        self.cost = lambda *args, **kw: np.nan
         self.time_dimension = time_dimension
 
     def _op_method(self, data, **kwargs):

src/patch_denoise/bindings/utils.py

@@ -1,6 +1,9 @@
 """Common utilities for bindings."""
-from dataclasses import dataclass
+
+from __future__ import annotations
+
 import logging
+from dataclasses import dataclass
 
 import numpy as np
 

src/patch_denoise/space_time/base.py

@@ -1,7 +1,8 @@
 """Base Structure for patch-based denoising on spatio-temporal dimension."""
+
 import abc
-from functools import partial, cached_property
 import logging
+
 import numpy as np
 from tqdm.auto import tqdm
 
@@ -43,7 +44,8 @@ def __init__(
 
         if self._ps.size != dimensions or step.size != dimensions:
             raise ValueError(
-                "self._ps and step must have the same number of dimensions as the input self._array."
+                "self._ps and step must have the same number of dimensions as the "
+                "input self._array."
             )
 
         # Ensure patch size is not larger than self._array size along each axis

src/patch_denoise/space_time/lowrank.py

@@ -1,4 +1,5 @@
-"""Low Rank  methods."""
+"""Low Rank methods."""
+
 from types import MappingProxyType
 
 import numpy as np
@@ -9,7 +10,7 @@
 from .utils import (
     eig_analysis,
     eig_synthesis,
-    marshenko_pastur_median,
+    marchenko_pastur_median,
     svd_analysis,
     svd_synthesis,
 )
@@ -25,7 +26,7 @@
 
 @fill_doc
 class MPPCADenoiser(BaseSpaceTimeDenoiser):
-    """Denoising using the MP-PCA threshoding.
+    """Denoising using Marchenko-Pastur principal components analysis thresholding.
 
     Parameters
     ----------
@@ -127,7 +128,7 @@ class RawSVDDenoiser(BaseSpaceTimeDenoiser):
     ----------
     $patch_config
     threshold_vlue: float
-        treshold value for the singular values.
+        threshold value for the singular values.
     """
 
     def __init__(
@@ -180,7 +181,7 @@ def _patch_processing(self, patch, patch_idx=None, **kwargs):
 
         # Equation (3) in Manjon 2013
 
-        return p_new, maxidx, np.NaN
+        return p_new, maxidx, np.nan
 
 
 @fill_doc
@@ -204,7 +205,7 @@ def denoise(
     ):
         """Denoise using the NORDIC method.
 
-        Along with the input data a noise stp map or value should be provided.
+        Along with the input data a noise std map or value should be provided.
 
         Parameters
         ----------
@@ -291,7 +292,7 @@ class OptimalSVDDenoiser(BaseSpaceTimeDenoiser):
     ----------
     $patch_config
     loss: str
-        The loss determines the choise of the optimal thresholding function
+        The loss determines the choice of the optimal thresholding function
         associated to it. The losses `"fro"`, `"nuc"` and `"op"` are supported,
         for the frobenius, nuclear and operator norm, respectively.
     """
@@ -335,7 +336,7 @@ def denoise(
         $mask_config
         $noise_std
         loss: str
-            The loss for which the optimal thresholding is perform.
+            The loss for which the optimal thresholding is performed.
         eps_marshenko_pastur: float
             The precision with which the optimal threshold is computed.
 
@@ -345,7 +346,7 @@ def denoise(
 
         Notes
         -----
-        Reimplement of the original Matlab code [#]_ in python.
+        Reimplementation of the original Matlab code [#]_ in python.
 
         References
         ----------
@@ -356,7 +357,7 @@ def denoise(
         """
         p_s, p_o = self._get_patch_param(input_data.shape)
 
-        self.input_denoising_kwargs["mp_median"] = marshenko_pastur_median(
+        self.input_denoising_kwargs["mp_median"] = marchenko_pastur_median(
             beta=input_data.shape[-1] / np.prod(p_s),
             eps=eps_marshenko_pastur,
         )
@@ -402,7 +403,7 @@ def _patch_processing(
             maxidx = 0
             p_new = np.zeros_like(patch) + p_tmean
 
-        return p_new, maxidx, np.NaN
+        return p_new, maxidx, np.nan
 
 
 def _sure_atn_cost(X, method, sing_vals, gamma, sigma=None, tau=None):
@@ -500,7 +501,7 @@ def sure_tau(tau, *args):
 
     if tau0 is None:
         tau0 = np.log(np.median(sing_vals))
-    cost_glob = np.Inf
+    cost_glob = np.inf
     for g in gamma0:
         res_opti = minimize(
             lambda x: _sure_atn_cost(
@@ -630,4 +631,4 @@ def _patch_processing(
             maxidx = 0
             p_new = np.zeros_like(patch) + p_tmean
 
-        return p_new, maxidx, np.NaN
+        return p_new, maxidx, np.nan

src/patch_denoise/space_time/utils.py

@@ -1,4 +1,5 @@
 """Utilities for space-time denoising."""
+
 import numpy as np
 from scipy.integrate import quad
 from scipy.linalg import eigh, svd
@@ -145,7 +146,7 @@ def eig_synthesis(data_centered, eig_vec, mean, max_val):
     return ((data_centered @ eig_vec) @ eig_vec.conj().T) + mean
 
 
-def marshenko_pastur_median(beta, eps=1e-7):
+def marchenko_pastur_median(beta, eps=1e-7):
     r"""Compute the median of the Marchenko-Pastur Distribution.
 
     Parameters
@@ -204,19 +205,38 @@ def mp_pdf(x):
 
 
 def estimate_noise(noise_sequence, block_size=1):
-    """Estimate the temporal noise standard deviation of a noise only sequence."""
+    """Estimate a noise map from a noise only sequence.
+
+    The noise map is the standard deviation of the noise in each patch.
+
+    Parameters
+    ----------
+    noise_sequence : np.ndarray of shape (X, Y, Z, T)
+        The noise-only data.
+    block_size : int
+        The size of the patch used to estimate the noise.
+
+    Returns
+    -------
+    np.ndarray of shape (X, Y, Z)
+        The estimated noise map.
+    """
     volume_shape = noise_sequence.shape[:-1]
     noise_map = np.empty(volume_shape)
     patch_shape = (block_size,) * len(volume_shape)
     patch_overlap = (block_size - 1,) * len(volume_shape)
 
     for patch_tl in get_patch_locs(patch_shape, patch_overlap, volume_shape):
+        # Get the index of voxels in the patch
         patch_slice = tuple(
             slice(ptl, ptl + ps) for ptl, ps in zip(patch_tl, patch_shape)
         )
+        # Identify the voxel in the center of the patch
         patch_center_img = tuple(
             slice(ptl + ps // 2, ptl + ps // 2 + 1)
             for ptl, ps in zip(patch_tl, patch_shape)
         )
+        # Set the value of the voxel in the center of the patch to the SD of
+        # the patch
         noise_map[patch_center_img] = np.std(noise_sequence[patch_slice])
     return noise_map

tests/test_bindings.py

@@ -1,5 +1,7 @@
 """Test for the binding module."""
+
 import os
+
 import numpy as np
 import numpy.testing as npt
 import pytest
@@ -16,7 +18,6 @@
 except ImportError as e:
     NIPYPE_AVAILABLE = False
 
-
 from patch_denoise.bindings.modopt import LLRDenoiserOperator
 from patch_denoise.bindings.nipype import PatchDenoise
 from patch_denoise.bindings.utils import DenoiseParameters

tests/test_spacetime_utils.py

@@ -8,7 +8,7 @@
     eig_analysis,
     eig_synthesis,
     estimate_noise,
-    marshenko_pastur_median,
+    marchenko_pastur_median,
     svd_analysis,
     svd_synthesis,
 )
@@ -55,7 +55,7 @@ def f(x):
         else:
             return 0
 
-    integral_median = marshenko_pastur_median(beta, eps=1e-7)
+    integral_median = marchenko_pastur_median(beta, eps=1e-7)
 
     vals = np.linspace(beta_m, beta_p, n_samples)
     proba = np.array(list(map(f, vals)))