Merge pull request #568 from Deep-MI/numpy2

Numpy2

FastSurferCNN/data_loader/conform.py

@@ -635,7 +635,8 @@ def conform(
     # Pxyz is the center of the image in world coords
 
     # target scalar type and dtype
-    sctype = np.uint8 if dtype is None else np.obj2sctype(dtype, default=np.uint8)
+    #sctype = np.uint8 if dtype is None else np.obj2sctype(dtype, default=np.uint8)
+    sctype = np.uint8 if dtype is None else np.dtype(dtype).type
     target_dtype = np.dtype(sctype)
 
     src_min, scale = 0, 1.0
@@ -761,7 +762,7 @@ def is_conform(
         raise ValueError(f"ERROR: Multiple input frames ({ishape[3]}) not supported!")
 
     checks = {
-        f"Number of Dimensions 3": (len(ishape) == 3, f"image ndim {img.ndim}")
+        "Number of Dimensions 3": (len(ishape) == 3, f"image ndim {img.ndim}")
     }
     # check dimensions
     if Criteria.FORCE_IMG_SIZE in criteria:
@@ -775,7 +776,7 @@ def is_conform(
     _vox_sizes = conformed_vox_size if is_correct_vox_size else izoom[:3]
     if Criteria.FORCE_ISO_VOX in criteria:
         vox_size_criteria = f"Voxel Size {'x'.join([str(conformed_vox_size)] * 3)}"
-        image_vox_size = f"image " + "x".join(map(str, izoom))
+        image_vox_size = "image " + "x".join(map(str, izoom))
         checks[vox_size_criteria] = (is_correct_vox_size, image_vox_size)
 
     # check orientation LIA
@@ -795,7 +796,8 @@ def is_conform(
         if dtype is None or (isinstance(dtype, str) and dtype.lower() == "uchar"):
             dtype = "uint8"
         else:  # assume obj
-            dtype = np.dtype(np.obj2sctype(dtype)).name
+            #dtype = np.dtype(np.obj2sctype(dtype)).name
+            dtype = np.dtype(dtype).type.__name__
         is_correct_dtype = img.get_data_dtype() == dtype
         checks[f"Dtype {dtype}"] = (is_correct_dtype, f"dtype {img.get_data_dtype()}")
 

Tutorial/Tutorial_FastSurferCNN_QuickSeg.ipynb

@@ -218,7 +218,7 @@
     "#@title Here we first setup the environment by downloading the open source deep-mi/fastsurfer project and the required packages\n",
     "import os\n",
     "import sys\n",
-    "from os.path import exists, join, basename, splitext\n",
+    "from os.path import exists, basename, splitext\n",
     "\n",
     "print(\"Starting setup. This could take a few minutes\")\n",
     "print(\"----------------------------------------------\")\n",
@@ -474,12 +474,12 @@
    "source": [
     "#@title Click this run button, if you would prefer to download the segmentation in nifti-format\n",
     "import nibabel as nib\n",
+    "from google.colab import files\n",
     "# conversion to nifti\n",
     "data = nib.load(f'{SETUP_DIR}fastsurfer_seg/Tutorial/mri/aparc.DKTatlas+aseg.deep.mgz')\n",
     "img_nifti = nib.Nifti1Image(data.get_fdata(), data.affine, header=nib.Nifti1Header())\n",
     "nib.nifti1.save(img_nifti, f'{SETUP_DIR}fastsurfer_seg/Tutorial/mri/aparc.DKTatlas+aseg.deep.nii.gz')\n",
     "\n",
-    "from google.colab import files\n",
     "files.download(f'{SETUP_DIR}fastsurfer_seg/Tutorial/mri/aparc.DKTatlas+aseg.deep.nii.gz')\n"
    ]
   },
@@ -519,12 +519,12 @@
    "source": [
     "#@title Click this run button, if you would prefer to download the image in nifti-format\n",
     "import nibabel as nib\n",
+    "from google.colab import files\n",
     "# conversion to nifti\n",
     "data = nib.load(f\"{SETUP_DIR}140_orig.mgz\")\n",
     "img_nifti = nib.Nifti1Image(data.get_fdata(), data.affine, header=nib.Nifti1Header())\n",
     "nib.nifti1.save(img_nifti, f\"{SETUP_DIR}140_orig.nii.gz\")\n",
     "\n",
-    "from google.colab import files\n",
     "files.download(f\"{SETUP_DIR}140_orig.nii.gz\")\n"
    ]
   },
@@ -612,11 +612,11 @@
     "%matplotlib inline\n",
     "import nibabel as nib\n",
     "import matplotlib.pyplot as plt\n",
-    "plt.style.use('seaborn-v0_8-whitegrid')\n",
-    "from skimage import color\n",
     "import torch\n",
     "import numpy as np\n",
+    "from skimage import color\n",
     "from torchvision import utils\n",
+    "plt.style.use('seaborn-v0_8-whitegrid')\n",
     "\n",
     "def plot_predictions(image, pred):\n",
     "    \"\"\"\n",
@@ -676,7 +676,6 @@
     "from ipywidgets import widgets\n",
     "import matplotlib.pyplot as plt\n",
     "import nibabel as nib\n",
-    "import numpy as np\n",
     "#from mpl_toolkits.mplot3d.art3d import Poly3DCollection\n",
     "from skimage import measure\n",
     "\n",
@@ -853,7 +852,7 @@
     "def plot_3d_plotly_shape(structure, hemisphere, show_mesh=True, crop=True, grid=True):\n",
     "  import plotly.graph_objects as go\n",
     "  label = label_lookups(structure, hemisphere)\n",
-    "  test_cond = np.in1d(pred_data, label).reshape(pred_data.shape)\n",
+    "  test_cond = np.isin(pred_data, label).reshape(pred_data.shape)\n",
     "  roi = np.where(test_cond, 1, 0)\n",
     "  vert_p, faces_p, normals_p, values_p = measure.marching_cubes(roi, 0, spacing=(1, 1, 1))\n",
     "\n",

recon_surf/N4_bias_correct.py

@@ -20,7 +20,7 @@
 import logging
 import sys
 from pathlib import Path
-from typing import Optional, cast, Literal, TypeVar, Callable, get_args, Union
+from typing import Optional, cast, Literal, TypeVar, Callable
 
 # Group 2: External modules
 import SimpleITK as sitk
@@ -499,9 +499,9 @@ def read_talairach_xfm(fname: Path | str) -> np.ndarray:
     try:
         transform_iter = iter(lines)
         # advance transform_iter to linear header
-        _ = next(l for l in transform_iter if l.lower().startswith("linear_"))
+        _ = next(ln for ln in transform_iter if ln.lower().startswith("linear_"))
         # return the next 3 lines in transform_lines
-        transform_lines = (l for l, _ in zip(transform_iter, range(3)))
+        transform_lines = (ln for ln, _ in zip(transform_iter, range(3)))
         tal_str = [ln.replace(";", " ") for ln in transform_lines]
         tal = np.genfromtxt(tal_str)
         tal = np.vstack([tal, [0, 0, 0, 1]])