suggested changes

wyli · Dec 10, 2020 · 0b5096e · 0b5096e
1 parent e932a88
commit 0b5096e
Showing 1 changed file with 60 additions and 24 deletions.
diff --git a/modules/CAM/class_lung_lesion.ipynb b/modules/CAM/class_lung_lesion.ipynb
@@ -46,28 +46,50 @@
         "id": "nZYowi8fUVp5"
       },
       "source": [
+    "import os\n",
         "import glob\n",
         "import random\n",
         "import numpy as np\n",
         "import torch\n",
         "from torch.utils.tensorboard import SummaryWriter\n",
+    "import matplotlib.pyplot as plt\n",
         "\n",
         "import monai\n",
         "from monai.transforms import (\n",
         "    AddChanneld, Compose, LoadImaged, RandRotate90d, \n",
         "    Resized, ScaleIntensityRanged, ToTensord, \n",
         "    RandFlipd, RandSpatialCropd,\n",
         ")\n",
+    "from monai.metrics import compute_occlusion_sensitivity\n",
         "\n",
         "monai.config.print_config()\n",
-        "monai.utils.set_determinism(42)\n",
-        "device = torch.device(\"cuda\" if torch.cuda.is_available() else \"cpu\")"
-      ],
-      "execution_count": null,
-      "outputs": []
-    },
-    {
-      "cell_type": "markdown",
+    "random_seed = 42\n",
+    "monai.utils.set_determinism(random_seed)\n",
+    "device = torch.device(\"cuda\" if torch.cuda.is_available() else \"cpu\")"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "use_patch_dataset=True  # switch this to use partial dataset or whole thing\n",
+    "\n",
+    "directory = os.environ.get(\"MONAI_DATA_DIRECTORY\")\n",
+    "root_dir = tempfile.mkdtemp() if directory is None else os.path.expanduser(directory)\n",
+    "data_path=os.path.join(root_dir, \"patch\")\n",
+    "if use_patch_dataset:\n",
+    "    !cd {root_dir} && [ -f \"lung_lesion_patches.tar.gz\" ] || gdown --id 1Jte6L7B_5q9XMgOCAq1Ldn29F1aaIxjW \\\n",
+    "        && mkdir -p {data_path} && tar -xvf \"lung_lesion_patches.tar.gz\" -C {data_path} > /dev/null\n",
+    "else:\n",
+    "    # TODO: took too long to download data to fix this.\n",
+    "    monai.apps.DecathlonDataset(root_dir=root_dir, task=\"Task06_Lung\", section=\"training\", download=True)\n",
+    "    %run -i './bbox_gen.py {root_dir}'"
+   ]
+  },
+  {
+   "cell_type": "markdown",
       "metadata": {
         "id": "YcfTvuxyy9jX"
       },
@@ -85,8 +107,8 @@
         "outputId": "773bfa78-4a34-4692-9540-f6bb0babd7d9"
       },
       "source": [
-        "lesion = glob.glob(\"./patch/lesion_*\")\n",
-        "non_lesion = glob.glob(\"./patch/norm_*\")\n",
+    "lesion = glob.glob(os.path.join(data_path,\"lesion_*\"))\n",
+    "non_lesion = glob.glob(os.path.join(data_path,\"norm_*\"))\n",
         "labels = np.asarray([[0., 1.]] * len(lesion) + [[1., 0.]] * len(non_lesion))\n",
         "\n",
         "all_files = [{\"image\": img, \"label\": label} for img, label in zip(lesion + non_lesion, labels)]\n",
@@ -235,12 +257,16 @@
         "id": "rA_cp54ebxRv"
       },
       "source": [
+    "from IPython.display import clear_output\n",
+    "\n",
         "# start training\n",
         "val_interval, total_epochs = 1, 30\n",
         "best_metric = best_metric_epoch = -1\n",
-        "epoch_loss_values = metric_values = []\n",
+    "epoch_loss_values = []\n",
+    "metric_values = []\n",
         "scaler = torch.cuda.amp.GradScaler()\n",
         "for epoch in range(total_epochs):\n",
+    "    clear_output()\n",
         "    print(\"-\" * 10)\n",
         "    print(f\"epoch {epoch + 1}/{total_epochs}\")\n",
         "    model.train()\n",
@@ -284,15 +310,28 @@
         "            print(\n",
         "                \"current epoch: {} current accuracy: {:.4f} best accuracy: {:.4f} at epoch {}\".format(\n",
         "                    epoch + 1, metric, best_metric, best_metric_epoch\n",
-        "                )\n",
-        "            )\n",
-        "print(f\"train completed, best_metric: {best_metric:.4f} at epoch: {best_metric_epoch}\")"
-      ],
-      "execution_count": null,
-      "outputs": []
-    },
-    {
-      "cell_type": "markdown",
+    "                )\n",
+    "            )\n",
+    "print(f\"train completed, best_metric: {best_metric:.4f} at epoch: {best_metric_epoch}\")"
+   ]
+  },
+  {
+   "cell_type": "code",
+   "execution_count": null,
+   "metadata": {},
+   "outputs": [],
+   "source": [
+    "import matplotlib.pyplot as plt\n",
+    "plt.plot(epoch_loss_values,label='train')\n",
+    "val_epochs=np.linspace(1, total_epochs,np.floor(total_epochs/val_interval).astype(np.int32))\n",
+    "plt.plot(val_epochs, metric_values,label='validation')\n",
+    "plt.legend()\n",
+    "plt.xlabel('Epoch')\n",
+    "plt.ylabel('Loss')"
+   ]
+  },
+  {
+   "cell_type": "markdown",
       "metadata": {
         "id": "iHE0RMEFIera"
       },
@@ -341,10 +380,7 @@
         "outputId": "ffcda223-4c1a-42be-a0b6-7f393d06e25a"
       },
       "source": [
-        "import matplotlib.pyplot as plt\n",
-        "\n",
-        "monai.utils.set_determinism(42)\n",
-        "train_transforms.set_random_state(42)\n",
+    "train_transforms.set_random_state(random_seed)\n",
         "items = [41, 45, 80, 78, 10]\n",
         "plt.subplots(2, len(items))\n",
         "for idx, item in enumerate(items):\n",