From 5d822a49eb15087168d4c92a1a23acc963e8d26c Mon Sep 17 00:00:00 2001 From: Mingxin Zheng <18563433+mingxin-zheng@users.noreply.github.com> Date: Thu, 24 Nov 2022 17:33:32 +0800 Subject: [PATCH] fix plotting and correct typos Signed-off-by: Mingxin Zheng <18563433+mingxin-zheng@users.noreply.github.com> --- .../notebooks/auto3dseg_hello_world.ipynb | 407 +++--------------- 1 file changed, 50 insertions(+), 357 deletions(-) diff --git a/auto3dseg/notebooks/auto3dseg_hello_world.ipynb b/auto3dseg/notebooks/auto3dseg_hello_world.ipynb index 198cf7e75d..3bfc1034dc 100644 --- a/auto3dseg/notebooks/auto3dseg_hello_world.ipynb +++ b/auto3dseg/notebooks/auto3dseg_hello_world.ipynb @@ -6,48 +6,37 @@ "source": [ "# MONAI Auto3DSeg \"Hello World\" Example\n", "\n", - "In this notebook, we will provide a simple demonstration of how to use Auto3DSeg AutoRunner to process a simulated datasets and generate results within minutes.\n", - "\n", - "## 1. Set up environment, imports and datasets" + "In this notebook, we will provide a simple demonstration of how to use Auto3DSeg AutoRunner to process a simulated dataset and generate results within minutes." ] }, { "cell_type": "markdown", "metadata": {}, "source": [ - "### 1.1 Set up Environment" + "## Setup environment" ] }, { "cell_type": "code", - "execution_count": 1, + "execution_count": null, "metadata": {}, "outputs": [], "source": [ - "!python -c \"import monai\" || pip install -q \"monai-weekly[nibabel]\"" + "!python -c \"import monai\" || pip install -q \"monai-weekly[nibabel, yaml, tqdm]\"" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ - "### 1.2 Set up imports" + "## Setup imports" ] }, { "cell_type": "code", - "execution_count": 2, + "execution_count": null, "metadata": {}, - "outputs": [ - { - "name": "stderr", - "output_type": "stream", - "text": [ - "/opt/conda/lib/python3.8/site-packages/tqdm/auto.py:22: TqdmWarning: IProgress not found. Please update jupyter and ipywidgets. See https://ipywidgets.readthedocs.io/en/stable/user_install.html\n", - " from .autonotebook import tqdm as notebook_tqdm\n" - ] - } - ], + "outputs": [], "source": [ "import os\n", "import json\n", @@ -65,16 +54,16 @@ "cell_type": "markdown", "metadata": {}, "source": [ - "### 1.3 Simulate a special datasets\n", + "## Simulate a special dataset\n", "\n", - "It is well known that AI takes time to train. To provide the \"Hello World!\" experience of Auto3D in this notebook, we will simulate a small datasets and run trainings only for multiple epocs. Due to the nature of AI, the performance shouldn't be highly expected, but the entire pipeline will be completed within minutes!\n", + "It is well known that AI takes time to train. To provide the \"Hello World!\" experience of Auto3D in this notebook, we will simulate a small dataset and run training only for multiple epochs. Due to the nature of AI, the performance shouldn't be highly expected, but the entire pipeline will be completed within minutes!\n", "\n", "`sim_datalist` provides the information of the simulated datasets. It lists 12 training and 2 testing images and labels. The training data are split into 3 folds. Each fold will use 8 images to train and 4 images to validate. The size of the dimension is defined by the `sim_dim` ." ] }, { "cell_type": "code", - "execution_count": 3, + "execution_count": null, "metadata": {}, "outputs": [], "source": [ @@ -106,14 +95,14 @@ "cell_type": "markdown", "metadata": {}, "source": [ - "### 1.4 Generate images and labels\n", + "## Generate images and labels\n", "\n", "Now we can use MONAI `create_test_image_3d` and `nib.Nifti1Image` functions to generate the 3D simulated images under the work_dir" ] }, { "cell_type": "code", - "execution_count": 4, + "execution_count": null, "metadata": {}, "outputs": [], "source": [ @@ -143,26 +132,24 @@ "cell_type": "markdown", "metadata": {}, "source": [ - "### 1.5 Inspect the simulated data\n", + "## Inspect the simulated data\n", "\n", - "Here we picked a slice of the 3D image in the test datasets to get an idea of the simulation" + "Here we pick a slice of the 3D image in the test datasets to get an idea of the simulated images." ] }, { "cell_type": "code", - "execution_count": 5, + "execution_count": null, "metadata": {}, "outputs": [ { "data": { - "image/png": "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", + "image/png": "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", "text/plain": [ - "
" + "
" ] }, - "metadata": { - "needs_background": "light" - }, + "metadata": {}, "output_type": "display_data" } ], @@ -178,35 +165,25 @@ "plt.subplot(1, 2, 2)\n", "plt.imshow(lbl[32])\n", "plt.title('label')\n", - "cbar = plt.colorbar(shrink=0.55)\n" + "cbar = plt.colorbar(shrink=0.55)" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ - "## 2. Test the Auto3DSeg AutoRunner pipeline\n", + "## Test the Auto3DSeg AutoRunner pipeline\n", "\n", - "AutoRunner provides an interface to run the Auot3Dseg pipeline. It will automatically configure four different neural networks and perform multi-fold training to achieve the state-of-the-art performance. The module is highly configurable, but only requires minimal user inputs. In this example, we want to complete the piepline in two minutes, so we need to have some customizations\n", + "`AutoRunner` provides an interface to run the Auot3DSeg pipeline. It will automatically configure four different neural networks and perform multi-fold training to achieve state-of-the-art performance. The module is highly configurable and only requires minimal user input to begin with. In this example, we want to complete the pipeline in two minutes, so we need to have some customizations.\n", "\n", - "### 2.1 Set up the AutoRunner with default settings" + "Now we set up the `AutoRunner` with some settings:" ] }, { "cell_type": "code", - "execution_count": 6, + "execution_count": null, "metadata": {}, - "outputs": [ - { - "name": "stdout", - "output_type": "stream", - "text": [ - "2022-10-18 08:27:25,835 - INFO - Work directory helloworld_work_dir is used to save all results\n", - "2022-10-18 08:27:25,841 - INFO - The output_dir is not specified. /workspace/monai/tutorials-in-dev/auto3dseg/notebooks/helloworld_work_dir/ensemble_output will be used to save ensemble predictions\n", - "2022-10-18 08:27:25,841 - INFO - Directory /workspace/monai/tutorials-in-dev/auto3dseg/notebooks/helloworld_work_dir/ensemble_output is created to save ensemble predictions\n" - ] - } - ], + "outputs": [], "source": [ "runner = AutoRunner(\n", " work_dir=work_dir,\n", @@ -215,20 +192,19 @@ " \"datalist\": datalist_file,\n", " \"dataroot\": dataroot_dir,\n", " }\n", - ")\n", - "# runner.run() # this will use the default settings and take hours/ to run " + ")" ] }, { "cell_type": "markdown", "metadata": {}, "source": [ - "### 2.2 Override the training parameters so that we can complete the pipeline in minutes" + "## Override the training parameters so that we can complete the pipeline in minutes" ] }, { "cell_type": "code", - "execution_count": 7, + "execution_count": null, "metadata": {}, "outputs": [], "source": [ @@ -253,300 +229,14 @@ "cell_type": "markdown", "metadata": {}, "source": [ - "### 2.3 Start the data anlaysis, algorithm generation, training, and model ensemble" + "## Start the data analysis, algorithm generation, training, and model ensemble" ] }, { "cell_type": "code", - "execution_count": 8, + "execution_count": null, "metadata": {}, - "outputs": [ - { - "name": "stderr", - "output_type": "stream", - "text": [ - "100%|██████████| 12/12 [00:00<00:00, 32.40it/s]\n", - "algo_templates.tar.gz: 296kB [00:01, 257kB/s] " - ] - }, - { - "name": "stdout", - "output_type": "stream", - "text": [ - "2022-10-18 08:27:27,916 - INFO - Downloaded: /tmp/tmpkajbqsnz/algo_templates.tar.gz\n", - "2022-10-18 08:27:27,918 - INFO - Expected md5 is None, skip md5 check for file /tmp/tmpkajbqsnz/algo_templates.tar.gz.\n", - "2022-10-18 08:27:27,920 - INFO - Writing into directory: /workspace/monai/tutorials-in-dev/auto3dseg/notebooks/helloworld_work_dir.\n" - ] - }, - { - "name": "stderr", - "output_type": "stream", - "text": [ - "\n" - ] - }, - { - "name": "stdout", - "output_type": "stream", - "text": [ - "2022-10-18 08:27:28,111 - INFO - /workspace/monai/tutorials-in-dev/auto3dseg/notebooks/helloworld_work_dir/segresnet2d_0\n", - "2022-10-18 08:27:28,237 - INFO - /workspace/monai/tutorials-in-dev/auto3dseg/notebooks/helloworld_work_dir/dints_0\n", - "2022-10-18 08:27:28,351 - INFO - /workspace/monai/tutorials-in-dev/auto3dseg/notebooks/helloworld_work_dir/swinunetr_0\n", - "2022-10-18 08:27:28,509 - INFO - /workspace/monai/tutorials-in-dev/auto3dseg/notebooks/helloworld_work_dir/segresnet_0\n", - "2022-10-18 08:27:28,510 - INFO - Launching: python /workspace/monai/tutorials-in-dev/auto3dseg/notebooks/helloworld_work_dir/segresnet_0/scripts/train.py run --config_file='/workspace/monai/tutorials-in-dev/auto3dseg/notebooks/helloworld_work_dir/segresnet_0/configs/transforms_validate.yaml','/workspace/monai/tutorials-in-dev/auto3dseg/notebooks/helloworld_work_dir/segresnet_0/configs/transforms_train.yaml','/workspace/monai/tutorials-in-dev/auto3dseg/notebooks/helloworld_work_dir/segresnet_0/configs/transforms_infer.yaml','/workspace/monai/tutorials-in-dev/auto3dseg/notebooks/helloworld_work_dir/segresnet_0/configs/network.yaml','/workspace/monai/tutorials-in-dev/auto3dseg/notebooks/helloworld_work_dir/segresnet_0/configs/hyper_parameters.yaml' --num_iterations=8 --num_iterations_per_validation=4 --num_images_per_batch=2 --num_epochs=2 --num_warmup_iterations=4\n", - "2022-10-18 08:27:35,294 - INFO - CompletedProcess(args=['python', '/workspace/monai/tutorials-in-dev/auto3dseg/notebooks/helloworld_work_dir/segresnet_0/scripts/train.py', 'run', \"--config_file='/workspace/monai/tutorials-in-dev/auto3dseg/notebooks/helloworld_work_dir/segresnet_0/configs/transforms_validate.yaml','/workspace/monai/tutorials-in-dev/auto3dseg/notebooks/helloworld_work_dir/segresnet_0/configs/transforms_train.yaml','/workspace/monai/tutorials-in-dev/auto3dseg/notebooks/helloworld_work_dir/segresnet_0/configs/transforms_infer.yaml','/workspace/monai/tutorials-in-dev/auto3dseg/notebooks/helloworld_work_dir/segresnet_0/configs/network.yaml','/workspace/monai/tutorials-in-dev/auto3dseg/notebooks/helloworld_work_dir/segresnet_0/configs/hyper_parameters.yaml'\", '--num_iterations=8', '--num_iterations_per_validation=4', '--num_images_per_batch=2', '--num_epochs=2', '--num_warmup_iterations=4'], returncode=0, stdout=b\"[info] number of GPUs: 1\n", - "[info] world_size: 1\n", - "train_files: 8\n", - "val_files: 4\n", - "num_epochs 2\n", - "num_epochs_per_validation 1\n", - "[info] training from scratch\n", - "[info] amp enabled\n", - "----------\n", - "epoch 1/2\n", - "learning rate is set to 2e-05\n", - "[2022-10-18 08:27:32] 1/4, train_loss: 1.4624\n", - "[2022-10-18 08:27:33] 2/4, train_loss: 1.4597\n", - "[2022-10-18 08:27:33] 3/4, train_loss: 1.3316\n", - "[2022-10-18 08:27:33] 4/4, train_loss: 1.3588\n", - "epoch 1 average loss: 1.4031, best mean dice: -1.0000 at epoch -1\n", - "1 / 4 tensor([[0.1104]], device='cuda:0')\n", - "2 / 4 tensor([[0.1104]], device='cuda:0')\n", - "3 / 4 tensor([[0.1104]], device='cuda:0')\n", - "4 / 4 tensor([[0.1104]], device='cuda:0')\n", - "evaluation metric - class 1: 0.11038827896118164\n", - "avg_metric 0.11038827896118164\n", - "saved new best metric model\n", - "current epoch: 1 current mean dice: 0.1104 best mean dice: 0.1104 at epoch 1\n", - "----------\n", - "epoch 2/2\n", - "learning rate is set to 0.0\n", - "[2022-10-18 08:27:34] 1/4, train_loss: 1.3096\n", - "[2022-10-18 08:27:34] 2/4, train_loss: 1.3398\n", - "[2022-10-18 08:27:34] 3/4, train_loss: 1.2666\n", - "[2022-10-18 08:27:34] 4/4, train_loss: 1.2679\n", - "epoch 2 average loss: 1.2960, best mean dice: 0.1104 at epoch 1\n", - "1 / 4 tensor([[0.1236]], device='cuda:0')\n", - "2 / 4 tensor([[0.1236]], device='cuda:0')\n", - "3 / 4 tensor([[0.1236]], device='cuda:0')\n", - "4 / 4 tensor([[0.1236]], device='cuda:0')\n", - "evaluation metric - class 1: 0.12358943372964859\n", - "avg_metric 0.12358943372964859\n", - "saved new best metric model\n", - "current epoch: 2 current mean dice: 0.1236 best mean dice: 0.1236 at epoch 2\n", - "train completed, best_metric: 0.1236 at epoch: 2\n", - "0.12358943372964859\n", - "\", stderr=b'')\n", - "2022-10-18 08:27:35,301 - INFO - Launching: python /workspace/monai/tutorials-in-dev/auto3dseg/notebooks/helloworld_work_dir/swinunetr_0/scripts/train.py run --config_file='/workspace/monai/tutorials-in-dev/auto3dseg/notebooks/helloworld_work_dir/swinunetr_0/configs/transforms_validate.yaml','/workspace/monai/tutorials-in-dev/auto3dseg/notebooks/helloworld_work_dir/swinunetr_0/configs/transforms_train.yaml','/workspace/monai/tutorials-in-dev/auto3dseg/notebooks/helloworld_work_dir/swinunetr_0/configs/transforms_infer.yaml','/workspace/monai/tutorials-in-dev/auto3dseg/notebooks/helloworld_work_dir/swinunetr_0/configs/network.yaml','/workspace/monai/tutorials-in-dev/auto3dseg/notebooks/helloworld_work_dir/swinunetr_0/configs/hyper_parameters.yaml' --num_iterations=8 --num_iterations_per_validation=4 --num_images_per_batch=2 --num_epochs=2 --num_warmup_iterations=4\n", - "2022-10-18 08:27:42,721 - INFO - CompletedProcess(args=['python', '/workspace/monai/tutorials-in-dev/auto3dseg/notebooks/helloworld_work_dir/swinunetr_0/scripts/train.py', 'run', \"--config_file='/workspace/monai/tutorials-in-dev/auto3dseg/notebooks/helloworld_work_dir/swinunetr_0/configs/transforms_validate.yaml','/workspace/monai/tutorials-in-dev/auto3dseg/notebooks/helloworld_work_dir/swinunetr_0/configs/transforms_train.yaml','/workspace/monai/tutorials-in-dev/auto3dseg/notebooks/helloworld_work_dir/swinunetr_0/configs/transforms_infer.yaml','/workspace/monai/tutorials-in-dev/auto3dseg/notebooks/helloworld_work_dir/swinunetr_0/configs/network.yaml','/workspace/monai/tutorials-in-dev/auto3dseg/notebooks/helloworld_work_dir/swinunetr_0/configs/hyper_parameters.yaml'\", '--num_iterations=8', '--num_iterations_per_validation=4', '--num_images_per_batch=2', '--num_epochs=2', '--num_warmup_iterations=4'], returncode=0, stdout=b\"[info] number of GPUs: 1\n", - "[info] world_size: 1\n", - "train_files: 8\n", - "val_files: 4\n", - "num_epochs 2\n", - "num_epochs_per_validation 1\n", - "[info] training from scratch\n", - "[info] amp enabled\n", - "----------\n", - "epoch 1/2\n", - "learning rate is set to 0.0001\n", - "[2022-10-18 08:27:39] 1/4, train_loss: 1.1682\n", - "[2022-10-18 08:27:39] 2/4, train_loss: 1.0048\n", - "[2022-10-18 08:27:39] 3/4, train_loss: 0.9436\n", - "[2022-10-18 08:27:40] 4/4, train_loss: 0.9151\n", - "epoch 1 average loss: 1.0079, best mean dice: -1.0000 at epoch -1\n", - "1 / 4 tensor([[0.6443]], device='cuda:0')\n", - "2 / 4 tensor([[0.6443]], device='cuda:0')\n", - "3 / 4 tensor([[0.6443]], device='cuda:0')\n", - "4 / 4 tensor([[0.6443]], device='cuda:0')\n", - "evaluation metric - class 1: 0.6442577242851257\n", - "avg_metric 0.6442577242851257\n", - "saved new best metric model\n", - "current epoch: 1 current mean dice: 0.6443 best mean dice: 0.6443 at epoch 1\n", - "----------\n", - "epoch 2/2\n", - "learning rate is set to 6.25e-06\n", - "[2022-10-18 08:27:41] 1/4, train_loss: 0.9040\n", - "[2022-10-18 08:27:41] 2/4, train_loss: 0.8981\n", - "[2022-10-18 08:27:41] 3/4, train_loss: 0.8954\n", - "[2022-10-18 08:27:41] 4/4, train_loss: 0.8919\n", - "epoch 2 average loss: 0.8974, best mean dice: 0.6443 at epoch 1\n", - "1 / 4 tensor([[0.6603]], device='cuda:0')\n", - "2 / 4 tensor([[0.6603]], device='cuda:0')\n", - "3 / 4 tensor([[0.6603]], device='cuda:0')\n", - "4 / 4 tensor([[0.6603]], device='cuda:0')\n", - "evaluation metric - class 1: 0.6603014469146729\n", - "avg_metric 0.6603014469146729\n", - "saved new best metric model\n", - "current epoch: 2 current mean dice: 0.6603 best mean dice: 0.6603 at epoch 2\n", - "train completed, best_metric: 0.6603 at epoch: 2\n", - "0.6603014469146729\n", - "\", stderr=b'')\n", - "2022-10-18 08:27:42,728 - INFO - Launching: python /workspace/monai/tutorials-in-dev/auto3dseg/notebooks/helloworld_work_dir/dints_0/scripts/search.py run --config_file='/workspace/monai/tutorials-in-dev/auto3dseg/notebooks/helloworld_work_dir/dints_0/configs/transforms_validate.yaml','/workspace/monai/tutorials-in-dev/auto3dseg/notebooks/helloworld_work_dir/dints_0/configs/transforms_train.yaml','/workspace/monai/tutorials-in-dev/auto3dseg/notebooks/helloworld_work_dir/dints_0/configs/transforms_infer.yaml','/workspace/monai/tutorials-in-dev/auto3dseg/notebooks/helloworld_work_dir/dints_0/configs/hyper_parameters_search.yaml','/workspace/monai/tutorials-in-dev/auto3dseg/notebooks/helloworld_work_dir/dints_0/configs/network.yaml','/workspace/monai/tutorials-in-dev/auto3dseg/notebooks/helloworld_work_dir/dints_0/configs/network_search.yaml','/workspace/monai/tutorials-in-dev/auto3dseg/notebooks/helloworld_work_dir/dints_0/configs/hyper_parameters.yaml' --searching#num_iterations=8 --searching#num_iterations_per_validation=4 --searching#num_images_per_batch=2 --searching#num_epochs=2 --searching#num_warmup_iterations=4\n", - "2022-10-18 08:28:11,311 - INFO - CompletedProcess(args=['python', '/workspace/monai/tutorials-in-dev/auto3dseg/notebooks/helloworld_work_dir/dints_0/scripts/search.py', 'run', \"--config_file='/workspace/monai/tutorials-in-dev/auto3dseg/notebooks/helloworld_work_dir/dints_0/configs/transforms_validate.yaml','/workspace/monai/tutorials-in-dev/auto3dseg/notebooks/helloworld_work_dir/dints_0/configs/transforms_train.yaml','/workspace/monai/tutorials-in-dev/auto3dseg/notebooks/helloworld_work_dir/dints_0/configs/transforms_infer.yaml','/workspace/monai/tutorials-in-dev/auto3dseg/notebooks/helloworld_work_dir/dints_0/configs/hyper_parameters_search.yaml','/workspace/monai/tutorials-in-dev/auto3dseg/notebooks/helloworld_work_dir/dints_0/configs/network.yaml','/workspace/monai/tutorials-in-dev/auto3dseg/notebooks/helloworld_work_dir/dints_0/configs/network_search.yaml','/workspace/monai/tutorials-in-dev/auto3dseg/notebooks/helloworld_work_dir/dints_0/configs/hyper_parameters.yaml'\", '--searching#num_iterations=8', '--searching#num_iterations_per_validation=4', '--searching#num_images_per_batch=2', '--searching#num_epochs=2', '--searching#num_warmup_iterations=4'], returncode=0, stdout=b\"[info] number of GPUs: 1\n", - "[info] world_size: 1\n", - "train_files_w: 4\n", - "train_files_a: 4\n", - "val_files: 4\n", - "num_epochs 4\n", - "num_epochs_warmup 2\n", - "num_epochs_per_validation 2\n", - "[info] amp enabled\n", - "----------\n", - "epoch 1/4\n", - "learning rate is set to 0.025\n", - "[2022-10-18 08:27:48] 1/2, train_loss: 0.8553\n", - "[2022-10-18 08:27:49] 2/2, train_loss: 0.8300\n", - "epoch 1 average loss: 0.8426, best mean dice: -1.0000 at epoch -1\n", - "----------\n", - "epoch 2/4\n", - "learning rate is set to 0.025\n", - "[2022-10-18 08:27:50] 1/2, train_loss: 0.8394\n", - "[2022-10-18 08:27:50] 2/2, train_loss: 0.8302\n", - "epoch 2 average loss: 0.8348, best mean dice: -1.0000 at epoch -1\n", - "1 / 4 tensor([[0.1750]], device='cuda:0')\n", - "2 / 4 tensor([[0.1750]], device='cuda:0')\n", - "3 / 4 tensor([[0.1750]], device='cuda:0')\n", - "4 / 4 tensor([[0.1750]], device='cuda:0')\n", - "evaluation metric - class 1: 0.17498305439949036\n", - "avg_metric 0.17498305439949036\n", - "saved new best metric model\n", - "current epoch: 2 current mean dice: 0.1750 best mean dice: 0.1750 at epoch 2\n", - "----------\n", - "epoch 3/4\n", - "learning rate is set to 0.025\n", - "[2022-10-18 08:27:58] 1/2, train_loss: 0.8238\n", - "[2022-10-18 08:27:59] 1/2, train_loss_arch: 0.8115\n", - "[2022-10-18 08:28:00] 2/2, train_loss: 0.8093\n", - "[2022-10-18 08:28:00] 2/2, train_loss_arch: 0.8043\n", - "epoch 3 average loss: 0.8166, best mean dice: 0.1750 at epoch 2\n", - "epoch 3 average arch loss: 0.8079, best mean dice: 0.1750 at epoch 2\n", - "----------\n", - "epoch 4/4\n", - "learning rate is set to 0.00625\n", - "[2022-10-18 08:28:01] 1/2, train_loss: 0.7779\n", - "[2022-10-18 08:28:02] 1/2, train_loss_arch: 1.1365\n", - "[2022-10-18 08:28:03] 2/2, train_loss: 0.8037\n", - "[2022-10-18 08:28:04] 2/2, train_loss_arch: 1.1294\n", - "epoch 4 average loss: 0.7908, best mean dice: 0.1750 at epoch 2\n", - "epoch 4 average arch loss: 1.1330, best mean dice: 0.1750 at epoch 2\n", - "1 / 4 tensor([[0.1844]], device='cuda:0')\n", - "2 / 4 tensor([[0.1844]], device='cuda:0')\n", - "3 / 4 tensor([[0.1844]], device='cuda:0')\n", - "4 / 4 tensor([[0.1844]], device='cuda:0')\n", - "evaluation metric - class 1: 0.1843988001346588\n", - "avg_metric 0.1843988001346588\n", - "saved new best metric model\n", - "current epoch: 4 current mean dice: 0.1844 best mean dice: 0.1844 at epoch 4\n", - "train completed, best_metric: 0.1844 at epoch: 4\n", - "\", stderr=b'')\n", - "2022-10-18 08:28:11,312 - INFO - Launching: python /workspace/monai/tutorials-in-dev/auto3dseg/notebooks/helloworld_work_dir/dints_0/scripts/train.py run --config_file='/workspace/monai/tutorials-in-dev/auto3dseg/notebooks/helloworld_work_dir/dints_0/configs/transforms_validate.yaml','/workspace/monai/tutorials-in-dev/auto3dseg/notebooks/helloworld_work_dir/dints_0/configs/transforms_train.yaml','/workspace/monai/tutorials-in-dev/auto3dseg/notebooks/helloworld_work_dir/dints_0/configs/transforms_infer.yaml','/workspace/monai/tutorials-in-dev/auto3dseg/notebooks/helloworld_work_dir/dints_0/configs/hyper_parameters_search.yaml','/workspace/monai/tutorials-in-dev/auto3dseg/notebooks/helloworld_work_dir/dints_0/configs/network.yaml','/workspace/monai/tutorials-in-dev/auto3dseg/notebooks/helloworld_work_dir/dints_0/configs/network_search.yaml','/workspace/monai/tutorials-in-dev/auto3dseg/notebooks/helloworld_work_dir/dints_0/configs/hyper_parameters.yaml' --training#num_iterations=8 --training#num_iterations_per_validation=4 --training#num_images_per_batch=2 --training#num_epochs=2 --training#num_warmup_iterations=4\n", - "2022-10-18 08:28:21,854 - INFO - CompletedProcess(args=['python', '/workspace/monai/tutorials-in-dev/auto3dseg/notebooks/helloworld_work_dir/dints_0/scripts/train.py', 'run', \"--config_file='/workspace/monai/tutorials-in-dev/auto3dseg/notebooks/helloworld_work_dir/dints_0/configs/transforms_validate.yaml','/workspace/monai/tutorials-in-dev/auto3dseg/notebooks/helloworld_work_dir/dints_0/configs/transforms_train.yaml','/workspace/monai/tutorials-in-dev/auto3dseg/notebooks/helloworld_work_dir/dints_0/configs/transforms_infer.yaml','/workspace/monai/tutorials-in-dev/auto3dseg/notebooks/helloworld_work_dir/dints_0/configs/hyper_parameters_search.yaml','/workspace/monai/tutorials-in-dev/auto3dseg/notebooks/helloworld_work_dir/dints_0/configs/network.yaml','/workspace/monai/tutorials-in-dev/auto3dseg/notebooks/helloworld_work_dir/dints_0/configs/network_search.yaml','/workspace/monai/tutorials-in-dev/auto3dseg/notebooks/helloworld_work_dir/dints_0/configs/hyper_parameters.yaml'\", '--training#num_iterations=8', '--training#num_iterations_per_validation=4', '--training#num_images_per_batch=2', '--training#num_epochs=2', '--training#num_warmup_iterations=4'], returncode=0, stdout=b\"[info] number of GPUs: 1\n", - "[info] world_size: 1\n", - "train_files: 8\n", - "val_files: 4\n", - "num_epochs 2\n", - "num_epochs_per_validation 1\n", - "[info] training from scratch\n", - "[info] amp enabled\n", - "----------\n", - "epoch 1/2\n", - "learning rate is set to 0.2\n", - "[2022-10-18 08:28:16] 1/4, train_loss: 0.7010\n", - "[2022-10-18 08:28:17] 2/4, train_loss: 0.6845\n", - "[2022-10-18 08:28:17] 3/4, train_loss: 0.6449\n", - "[2022-10-18 08:28:17] 4/4, train_loss: 0.6553\n", - "epoch 1 average loss: 0.6714, best mean dice: -1.0000 at epoch -1\n", - "1 / 4 tensor([[0.2394]], device='cuda:0')\n", - "2 / 4 tensor([[0.2394]], device='cuda:0')\n", - "3 / 4 tensor([[0.2394]], device='cuda:0')\n", - "4 / 4 tensor([[0.2394]], device='cuda:0')\n", - "evaluation metric - class 1: 0.2393515557050705\n", - "avg_metric 0.2393515557050705\n", - "saved new best metric model\n", - "current epoch: 1 current mean dice: 0.2394 best mean dice: 0.2394 at epoch 1\n", - "----------\n", - "epoch 2/2\n", - "learning rate is set to 0.0125\n", - "[2022-10-18 08:28:19] 1/4, train_loss: 0.6216\n", - "[2022-10-18 08:28:19] 2/4, train_loss: 0.5904\n", - "[2022-10-18 08:28:19] 3/4, train_loss: 0.6117\n", - "[2022-10-18 08:28:20] 4/4, train_loss: 0.6234\n", - "epoch 2 average loss: 0.6118, best mean dice: 0.2394 at epoch 1\n", - "1 / 4 tensor([[0.2492]], device='cuda:0')\n", - "2 / 4 tensor([[0.2492]], device='cuda:0')\n", - "3 / 4 tensor([[0.2492]], device='cuda:0')\n", - "4 / 4 tensor([[0.2492]], device='cuda:0')\n", - "evaluation metric - class 1: 0.24918213486671448\n", - "avg_metric 0.24918213486671448\n", - "saved new best metric model\n", - "current epoch: 2 current mean dice: 0.2492 best mean dice: 0.2492 at epoch 2\n", - "train completed, best_metric: 0.2492 at epoch: 2\n", - "0.24918213486671448\n", - "\", stderr=b'')\n", - "2022-10-18 08:28:21,861 - INFO - Launching: python /workspace/monai/tutorials-in-dev/auto3dseg/notebooks/helloworld_work_dir/segresnet2d_0/scripts/train.py run --config_file='/workspace/monai/tutorials-in-dev/auto3dseg/notebooks/helloworld_work_dir/segresnet2d_0/configs/transforms_validate.yaml','/workspace/monai/tutorials-in-dev/auto3dseg/notebooks/helloworld_work_dir/segresnet2d_0/configs/transforms_train.yaml','/workspace/monai/tutorials-in-dev/auto3dseg/notebooks/helloworld_work_dir/segresnet2d_0/configs/transforms_infer.yaml','/workspace/monai/tutorials-in-dev/auto3dseg/notebooks/helloworld_work_dir/segresnet2d_0/configs/network.yaml','/workspace/monai/tutorials-in-dev/auto3dseg/notebooks/helloworld_work_dir/segresnet2d_0/configs/hyper_parameters.yaml' --num_iterations=8 --num_iterations_per_validation=4 --num_images_per_batch=2 --num_epochs=2 --num_warmup_iterations=4\n", - "2022-10-18 08:28:28,958 - INFO - CompletedProcess(args=['python', '/workspace/monai/tutorials-in-dev/auto3dseg/notebooks/helloworld_work_dir/segresnet2d_0/scripts/train.py', 'run', \"--config_file='/workspace/monai/tutorials-in-dev/auto3dseg/notebooks/helloworld_work_dir/segresnet2d_0/configs/transforms_validate.yaml','/workspace/monai/tutorials-in-dev/auto3dseg/notebooks/helloworld_work_dir/segresnet2d_0/configs/transforms_train.yaml','/workspace/monai/tutorials-in-dev/auto3dseg/notebooks/helloworld_work_dir/segresnet2d_0/configs/transforms_infer.yaml','/workspace/monai/tutorials-in-dev/auto3dseg/notebooks/helloworld_work_dir/segresnet2d_0/configs/network.yaml','/workspace/monai/tutorials-in-dev/auto3dseg/notebooks/helloworld_work_dir/segresnet2d_0/configs/hyper_parameters.yaml'\", '--num_iterations=8', '--num_iterations_per_validation=4', '--num_images_per_batch=2', '--num_epochs=2', '--num_warmup_iterations=4'], returncode=0, stdout=b\"[info] number of GPUs: 1\n", - "[info] world_size: 1\n", - "train_files: 8\n", - "val_files: 4\n", - "num_epochs 2\n", - "num_epochs_per_validation 1\n", - "[info] training from scratch\n", - "[info] amp enabled\n", - "----------\n", - "epoch 1/2\n", - "learning rate is set to 0.2\n", - "[2022-10-18 08:28:25] 1/4, train_loss: 0.6673\n", - "[2022-10-18 08:28:25] 2/4, train_loss: 0.5865\n", - "[2022-10-18 08:28:26] 3/4, train_loss: 0.3893\n", - "[2022-10-18 08:28:26] 4/4, train_loss: 0.3853\n", - "epoch 1 average loss: 0.5071, best mean dice: -1.0000 at epoch -1\n", - "1 / 4 tensor([[0.5780]], device='cuda:0')\n", - "2 / 4 tensor([[0.5780]], device='cuda:0')\n", - "3 / 4 tensor([[0.5780]], device='cuda:0')\n", - "4 / 4 tensor([[0.5780]], device='cuda:0')\n", - "evaluation metric - class 1: 0.5780409574508667\n", - "avg_metric 0.5780409574508667\n", - "saved new best metric model\n", - "current epoch: 1 current mean dice: 0.5780 best mean dice: 0.5780 at epoch 1\n", - "----------\n", - "epoch 2/2\n", - "learning rate is set to 0.0125\n", - "[2022-10-18 08:28:27] 1/4, train_loss: 0.4148\n", - "[2022-10-18 08:28:27] 2/4, train_loss: 0.3646\n", - "[2022-10-18 08:28:27] 3/4, train_loss: 0.6224\n", - "[2022-10-18 08:28:27] 4/4, train_loss: 0.4244\n", - "epoch 2 average loss: 0.4565, best mean dice: 0.5780 at epoch 1\n", - "1 / 4 tensor([[0.5818]], device='cuda:0')\n", - "2 / 4 tensor([[0.5818]], device='cuda:0')\n", - "3 / 4 tensor([[0.5818]], device='cuda:0')\n", - "4 / 4 tensor([[0.5818]], device='cuda:0')\n", - "evaluation metric - class 1: 0.581782341003418\n", - "avg_metric 0.581782341003418\n", - "saved new best metric model\n", - "current epoch: 2 current mean dice: 0.5818 best mean dice: 0.5818 at epoch 2\n", - "train completed, best_metric: 0.5818 at epoch: 2\n", - "0.581782341003418\n", - "\", stderr=b'')\n", - "0\n", - "[info] checkpoint /workspace/monai/tutorials-in-dev/auto3dseg/notebooks/helloworld_work_dir/segresnet_0/model_fold0/best_metric_model.pt loaded\n", - "2022-10-18 08:28:30,541 INFO image_writer.py:193 - writing: /workspace/monai/tutorials-in-dev/auto3dseg/notebooks/helloworld_work_dir/segresnet_0/prediction_testing/test_image_001/test_image_001_seg.nii.gz\n", - "[info] checkpoint /workspace/monai/tutorials-in-dev/auto3dseg/notebooks/helloworld_work_dir/segresnet2d_0/model_fold0/best_metric_model.pt loaded\n", - "2022-10-18 08:28:30,888 INFO image_writer.py:193 - writing: /workspace/monai/tutorials-in-dev/auto3dseg/notebooks/helloworld_work_dir/segresnet2d_0/prediction_testing/test_image_001/test_image_001_seg.nii.gz\n", - "1\n", - "[info] checkpoint /workspace/monai/tutorials-in-dev/auto3dseg/notebooks/helloworld_work_dir/segresnet_0/model_fold0/best_metric_model.pt loaded\n", - "2022-10-18 08:28:31,645 INFO image_writer.py:193 - writing: /workspace/monai/tutorials-in-dev/auto3dseg/notebooks/helloworld_work_dir/segresnet_0/prediction_testing/test_image_002/test_image_002_seg.nii.gz\n", - "[info] checkpoint /workspace/monai/tutorials-in-dev/auto3dseg/notebooks/helloworld_work_dir/segresnet2d_0/model_fold0/best_metric_model.pt loaded\n", - "2022-10-18 08:28:32,019 INFO image_writer.py:193 - writing: /workspace/monai/tutorials-in-dev/auto3dseg/notebooks/helloworld_work_dir/segresnet2d_0/prediction_testing/test_image_002/test_image_002_seg.nii.gz\n", - "Auto3Dseg picked the following networks to ensemble:\n", - "segresnet_0\n", - "segresnet2d_0\n", - "2022-10-18 08:28:32,025 INFO image_writer.py:193 - writing: /workspace/monai/tutorials-in-dev/auto3dseg/notebooks/helloworld_work_dir/ensemble_output/test_image_001/test_image_001_ensemble.nii.gz\n", - "2022-10-18 08:28:32,032 INFO image_writer.py:193 - writing: /workspace/monai/tutorials-in-dev/auto3dseg/notebooks/helloworld_work_dir/ensemble_output/test_image_002/test_image_002_ensemble.nii.gz\n", - "2022-10-18 08:28:32,038 - INFO - Auto3Dseg ensemble prediction outputs are saved in /workspace/monai/tutorials-in-dev/auto3dseg/notebooks/helloworld_work_dir/ensemble_output.\n", - "2022-10-18 08:28:32,039 - INFO - Auto3Dseg pipeline is complete successfully.\n" - ] - } - ], + "outputs": [], "source": [ "runner.run()" ] @@ -555,24 +245,22 @@ "cell_type": "markdown", "metadata": {}, "source": [ - "### 2.4 Inspect the algo ensembled predctions and compare it with the ground truth" + "## Inspect the algo ensembled predctions and compare it with the ground truth" ] }, { "cell_type": "code", - "execution_count": 9, + "execution_count": null, "metadata": {}, "outputs": [ { "data": { - "image/png": "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", + "image/png": "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", "text/plain": [ - "
" + "
" ] }, - "metadata": { - "needs_background": "light" - }, + "metadata": {}, "output_type": "display_data" } ], @@ -581,21 +269,26 @@ "prediction_nib = nib.load(os.path.join(work_dir, 'ensemble_output', image_name, image_name+\"_ensemble\"+\".nii.gz\"))\n", "pred = np.array(prediction_nib.dataobj)\n", "\n", + "img_slice32 = lbl[32] == 0\n", + "label_slice32 = lbl[32] == 1\n", + "background_slice32 = pred[32, :, :, 0] if pred.ndim == 4 else pred[32] == 0\n", + "foreground_slice32 = pred[32, :, :, 1] if pred.ndim == 4 else pred[32] == 1\n", + "\n", "plt.subplot(2, 2, 1)\n", - "plt.imshow(img[32])\n", - "plt.title('image')\n", + "plt.imshow(img_slice32)\n", + "plt.title('background groundtruth')\n", "cbar = plt.colorbar(shrink=0.8)\n", "plt.subplot(2, 2, 2)\n", - "plt.imshow(lbl[32])\n", - "plt.title('label')\n", + "plt.imshow(label_slice32)\n", + "plt.title('foreground_groundtruth')\n", "cbar = plt.colorbar(shrink=0.8)\n", "plt.subplot(2, 2, 3)\n", - "plt.imshow(pred[32, :, :, 0])\n", - "plt.title('background class prediction')\n", + "plt.imshow(background_slice32)\n", + "plt.title('background prediction')\n", "cbar = plt.colorbar(shrink=0.8)\n", "plt.subplot(2, 2, 4)\n", - "plt.imshow(pred[32, :, :, 1])\n", - "plt.title('foreground class prediction')\n", + "plt.imshow(foreground_slice32)\n", + "plt.title('foreground prediction')\n", "cbar = plt.colorbar(shrink=0.8)\n", "# set the spacing between subplots\n", "plt.subplots_adjust(left=0.1,\n", @@ -611,9 +304,9 @@ "cell_type": "markdown", "metadata": {}, "source": [ - "## 3. Conclusion\n", + "## Conclusion\n", "\n", - "Here we complete the entire pipeline. As you may notice, we can see the algorithms start to learn the datasets and the prediction of background and foreground voxels. Auto3DSeg and AutoRunner are highly configurable. To get better results, you can increase the training time, apply different ensemble method, or use hyper-parameter optimization through the AutoRunner or Auto3DSeg module APIs.\n", + "Here we complete the entire pipeline. As you may notice, we can see the algorithms start to learn the datasets and the prediction of background and foreground voxels. Auto3DSeg and AutoRunner are highly configurable. To get better results, you can increase the training time, apply a different ensemble method, or use hyper-parameter optimization through the AutoRunner or Auto3DSeg module APIs.\n", "\n", "As a \"Hello World\" notebook, we will end the experiment with simulated datasets here." ]