2 nifti stream and patch transforms (#28)

* initial unit tests for 2d/3d unet * adding license info * Adding definitions for reading Nifti files in streams and stream transforms for selecting patches (windowing) * Remove tests * Renaming fix * Update arrayutils.py * Removed blank lines in comments * Added Dataset based Nifti reader, grid patch sampler, and transforms * Added example segmentation notebook * Cleanup deletion * Update cardiac_segmentation.ipynb Co-authored-by: Wenqi Li <wenqi.li@ucl.ac.uk>
Project-MONAI · Jan 22, 2020 · 2fd7d6d · 2fd7d6d
1 parent eef2941
commit 2fd7d6d
Show file tree

Hide file tree

Showing 24 changed files with 844 additions and 1,732 deletions.
diff --git a/examples/cardiac_segmentation.ipynb b/examples/cardiac_segmentation.ipynb
diff --git a/examples/nifti_read_example.ipynb b/examples/nifti_read_example.ipynb
@@ -0,0 +1,233 @@
+{
+ "cells": [
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "# Nifti Read Example\n",
+    "\n",
+    "The purpose of this notebook is to illustrate reading Nifti files and iterating over patches of the volumes loaded from them."
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 1,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "MONAI version: 0.0.1\n",
+      "Python version: 3.7.3 (default, Mar 27 2019, 22:11:17)  [GCC 7.3.0]\n",
+      "Numpy version: 1.16.4\n",
+      "Pytorch version: 1.3.1\n",
+      "Ignite version: 0.2.1\n"
+     ]
+    }
+   ],
+   "source": [
+    "%matplotlib inline\n",
+    "\n",
+    "import os\n",
+    "import sys\n",
+    "from glob import glob\n",
+    "import tempfile\n",
+    "\n",
+    "import numpy as np\n",
+    "import matplotlib.pyplot as plt\n",
+    "import nibabel as nib\n",
+    "\n",
+    "\n",
+    "import torch\n",
+    "from torch.utils.data import DataLoader\n",
+    "import torchvision.transforms as transforms\n",
+    "\n",
+    "sys.path.append('..') # assumes this is where MONAI is\n",
+    "\n",
+    "from monai import application, data, networks, utils\n",
+    "from monai.data.readers import NiftiDataset\n",
+    "from monai.data.transforms import AddChannel, Transpose, Rescale, ToTensor, UniformRandomPatch, GridPatchDataset\n",
+    "\n",
+    "application.config.print_config()"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "Define a function for creating test images and segmentations:"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 2,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "def create_test_image_3d(height, width, depth, numObjs=12, radMax=30, noiseMax=0.0, numSegClasses=5):\n",
+    "    '''Return a noisy 3D image and segmentation.'''\n",
+    "    image = np.zeros((width, height,depth))\n",
+    "\n",
+    "    for i in range(numObjs):\n",
+    "        x = np.random.randint(radMax, width - radMax)\n",
+    "        y = np.random.randint(radMax, height - radMax)\n",
+    "        z = np.random.randint(radMax, depth - radMax)\n",
+    "        rad = np.random.randint(5, radMax)\n",
+    "        spy, spx, spz = np.ogrid[-x:width - x, -y:height - y, -z:depth - z]\n",
+    "        circle = (spx * spx + spy * spy + spz * spz) <= rad * rad\n",
+    "\n",
+    "        if numSegClasses > 1:\n",
+    "            image[circle] = np.ceil(np.random.random() * numSegClasses)\n",
+    "        else:\n",
+    "            image[circle] = np.random.random() * 0.5 + 0.5\n",
+    "\n",
+    "    labels = np.ceil(image).astype(np.int32)\n",
+    "\n",
+    "    norm = np.random.uniform(0, numSegClasses * noiseMax, size=image.shape)\n",
+    "    noisyimage = utils.arrayutils.rescale_array(np.maximum(image, norm))\n",
+    "\n",
+    "    return noisyimage, labels"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "Create a number of test Nifti files:"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 3,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "tempdir = tempfile.mkdtemp()\n",
+    "\n",
+    "for i in range(5):\n",
+    "    im, seg = create_test_image_3d(256,256,256)\n",
+    "    \n",
+    "    n = nib.Nifti1Image(im, np.eye(4))\n",
+    "    nib.save(n, os.path.join(tempdir, 'im%i.nii.gz'%i))\n",
+    "    \n",
+    "    n = nib.Nifti1Image(seg, np.eye(4))\n",
+    "    nib.save(n, os.path.join(tempdir, 'seg%i.nii.gz'%i))"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "Create a data loader which yields uniform random patches from loaded Nifti files:"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 4,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "torch.Size([5, 1, 64, 64, 64]) torch.Size([5, 1, 64, 64, 64])\n"
+     ]
+    }
+   ],
+   "source": [
+    "images = sorted(glob(os.path.join(tempdir,'im*.nii.gz')))\n",
+    "segs = sorted(glob(os.path.join(tempdir,'seg*.nii.gz')))\n",
+    "\n",
+    "imtrans=transforms.Compose([\n",
+    "    Rescale(),\n",
+    "    AddChannel(),\n",
+    "    UniformRandomPatch((64, 64, 64)),\n",
+    "    ToTensor()\n",
+    "])    \n",
+    "\n",
+    "segtrans=transforms.Compose([\n",
+    "    AddChannel(),\n",
+    "    UniformRandomPatch((64, 64, 64)),\n",
+    "    ToTensor()\n",
+    "])    \n",
+    "    \n",
+    "ds = NiftiDataset(images, segs, imtrans, segtrans)\n",
+    "\n",
+    "loader = DataLoader(ds, batch_size=10, num_workers=2, pin_memory=torch.cuda.is_available())\n",
+    "im, seg = utils.mathutils.first(loader)\n",
+    "print(im.shape, seg.shape)"
+   ]
+  },
+  {
+   "cell_type": "markdown",
+   "metadata": {},
+   "source": [
+    "Alternatively create a data loader which yields patches in regular grid order from loaded images:"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 5,
+   "metadata": {},
+   "outputs": [
+    {
+     "name": "stdout",
+     "output_type": "stream",
+     "text": [
+      "torch.Size([10, 1, 64, 64, 64]) torch.Size([10, 1, 64, 64, 64])\n"
+     ]
+    }
+   ],
+   "source": [
+    "imtrans=transforms.Compose([\n",
+    "    Rescale(),\n",
+    "    AddChannel(),\n",
+    "    ToTensor()\n",
+    "])    \n",
+    "\n",
+    "segtrans=transforms.Compose([\n",
+    "    AddChannel(),\n",
+    "    ToTensor()\n",
+    "])    \n",
+    "    \n",
+    "ds = NiftiDataset(images, segs, imtrans, segtrans)\n",
+    "ds = GridPatchDataset(ds, (64, 64, 64))\n",
+    "\n",
+    "loader = DataLoader(ds, batch_size=10, num_workers=2, pin_memory=torch.cuda.is_available())\n",
+    "im, seg = utils.mathutils.first(loader)\n",
+    "print(im.shape, seg.shape)"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": 6,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "!rm -rf {tempdir}"
+   ]
+  }
+ ],
+ "metadata": {
+  "kernelspec": {
+   "display_name": "Python 3",
+   "language": "python",
+   "name": "python3"
+  },
+  "language_info": {
+   "codemirror_mode": {
+    "name": "ipython",
+    "version": 3
+   },
+   "file_extension": ".py",
+   "mimetype": "text/x-python",
+   "name": "python",
+   "nbconvert_exporter": "python",
+   "pygments_lexer": "ipython3",
+   "version": "3.7.3"
+  }
+ },
+ "nbformat": 4,
+ "nbformat_minor": 4
+}