These setup instructions describe how to access the AWS accounts for the third day of the MONAI MICCAI Bootcamp 2021. Each account will be shared by two users and every user will launch a dedicated GPU-based SageMaker notebook instance to work on the ML challenges.
- Go to https://dashboard.eventengine.run/login and enter the event hash you have received via email and click on Accept Terms & Login:
- On the next page, click on Email One-Time Password (OTP):
- Enter the email address, which you used when you registered for the bootcamp. Using another email address won't allow you to access your temporary AWS account.
- In the next window, click on AWS Console:
- Another window will open. Please click on Open AWS Console:
- Congratulations, you have accessed your AWS account, which you will use for the rest of the day. Now, we need to make sure to launch your own Jupyter server.
- In the search bar on the top, please type "service catalog" and click on Service Catalog in the search result list:
- In Service Catalog, click on Bootcamp Notebook:
- Click on Launch product:
- Enter your specific notebook name into the Provisioned product name field and the NotebookInstanceName field. Remember, two users will share the same AWS account. Therefore, we recommend using a naming schema like
notebook-[firstnamelastname]
. The name must start with a letter (A-Z, a-z) or number (0-9) and can include hyphens (-), but not spaces. After that, click on Launch product at the bottom of the page (The launch button is not shown in the screenshot below). This will launch your pre-configured GPU-based SageMaker Jupyter notebook server in the background.
- After that, you will see a notification on the top of the page that your notebook is being launched:
- Now, we can leave Service Catalog.In the search bar at the top, enter "sagemaker" and click on SageMaker in the search result list:
- In the SageMaker Console, click on Notebook instances in the left sidebar:
- In the list of notebook instances, you will see the instance you just started via service catalog. If your instance is not yet in status "InService", just wait 1-2 minutes until the server was completely launched. Click on Open Jupyter Lab, which will open your Jupyter Lab environment in a separate tab.
- In Jupyter Lab, click on File > New > Terminal. We will now git clone the notebooks for today's challenge day.
- In the terminal, type
cd SageMaker
and press Enter. Then typegit clone https://github.com/Project-MONAI/MONAIBootcamp2021
and press Enter again. After that, you will be able to navigate to the notebooks you just downloaded via the file browser on the left:
- Congratulations! You are now ready to get started. The NVIDIA team will share specific information on the notebooks for the Bootcamp Challenge Day you will use today later during our intro session. But please, keep on reading the last section below.
Important: When selecting and launching a notebook, you will be asked to select your preferred kernel. Please make sure to always select the conda_pytorch-latest_p36 kernel
:
Have fun today and good luck with the mini challenges!
If you downloaded the notebooks before, please make sure to run the following instructions:
- In Jupyter Lab, click on File > New > Terminal. In the terminal, type
cd SageMaker
and press Enter. - Then type
git pull https://github.com/Project-MONAI/MONAIBootcamp2021
and press Enter again.
Below, we also added the links to the recourses, which were mentioned in the AWS on MONAI session on Day 2.
- Launching a Deep Learning AMI: https://aws.amazon.com/getting-started/hands-on/train-deep-learning-model-aws-ec2-containers/
- Launching a Deep Learning Container: https://aws.amazon.com/getting-started/hands-on/get-started-dlami/
- SageMaker Technical Deep Dive Series: https://www.youtube.com/watch?v=uQc8Itd4UTs&list=PLhr1KZpdzukcOr_6j_zmSrvYnLUtgqsZz
- Build a medical image analysis pipeline on Amazon SageMaker using the MONAI framework: https://aws.amazon.com/blogs/industries/build-a-medical-image-analysis-pipeline-on-amazon-sagemaker-using-the-monai-framework/
- GitHub repo: https://github.com/aws-samples/amazon-sagemaker-monai-examples
- Annotate DICOM images and build an ML model using the MONAI framework on Amazon SageMaker: https://aws.amazon.com/blogs/machine-learning/annotate-dicom-images-and-build-an-ml-model-using-the-monai-framework-on-amazon-sagemaker/
- GitHub repo: https://github.com/aws-samples/annotate-medical-images-in-dicom-server-and-build-ml-models-on-amazon-sagemaker
- Grand-challenge: https://grand-challenge.org
- MONAI dataset (Medical Segmentation Decathlon) on the Registry of Open Data: https://registry.opendata.aws/msd/