This put up is co-written with Stephen Aylward, Matt McCormick, Brianna Main from Kitware and Justin Kirby from the Frederick Nationwide Laboratory for Most cancers Analysis (FNLCR).
Amazon SageMaker Studio Lab gives no-cost access to a machine studying (ML) growth surroundings to everybody with an e mail tackle. Just like the totally featured Amazon SageMaker Studio, Studio Lab permits you to customise your personal Conda surroundings and create CPU- and GPU-scalable JupyterLab model 3 notebooks, with easy accessibility to the newest information science productiveness instruments and open-source libraries. Furthermore, Studio Lab free accounts embody a minimal of 15 GB of persistent storage, enabling you to repeatedly preserve and expend your initiatives throughout a number of periods and permitting you to immediately choose up the place your left off and even share your ongoing work and work environments with others.
A key difficulty confronted by the medical picture neighborhood is easy methods to allow researchers to experiment and discover with these important instruments. To unravel this problem, AWS groups labored with Kitware and Frederick National Laboratory for Cancer Research (FNLCR) to convey collectively three main medical imaging AI sources for Studio Lab and your entire open-source JupyterLab neighborhood:
These instruments and information mix to permit medical imaging AI researchers to shortly develop and completely consider clinically prepared deep studying algorithms in a complete and user-friendly surroundings. Staff members from FNLCR and Kitware collaborated to create a collection of Jupyter notebooks that show widespread workflows to programmatically entry and visualize TCIA information. These notebooks use Studio Lab to permit researchers to run the notebooks with out the necessity to arrange their very own native Jupyter growth surroundings—you possibly can shortly discover new concepts or combine your work into shows, workshops, and tutorials at conferences.
The next instance illustrates Studio Lab operating a Jupyter pocket book that downloads TCIA prostate MRI information, segments it utilizing MONAI, and shows the outcomes utilizing itkWidgets.
Though you possibly can simply perform smaller experiments and demos with the pattern notebooks introduced on this put up on Studio Lab at no cost, it is strongly recommended to make use of Amazon SageMaker Studio while you practice your personal medical picture fashions at scale. Amazon SageMaker Studio is an built-in web-based growth surroundings (IDE) with enterprise-grade safety, governance, and monitoring options from which you’ll entry purpose-built instruments to carry out all ML growth steps. Open-source libraries like MONAI Core and itkWidgets additionally run on Amazon SageMaker Studio.
Set up the answer
To run the TCIA notebooks on Studio Lab, you’ll want to register an account utilizing your e mail tackle on the Studio Lab website. Account requests might take 1–3 days to get accepted.
After that, you possibly can comply with the set up steps to get began:
- Log in to Studio Lab and begin a CPU runtime.
- In a separate tab, navigate to the TCIA notebooks GitHub repo and select a pocket book within the root folder of the repository.
- Select Open Studio Lab to open the pocket book in Studio Lab.
- Again in Studio Lab, select Copy to venture.
- Within the new JupyterLab pop-up that opens, select Clone Complete Repo.
- Within the subsequent window, hold the defaults and select Clone.
- Select OK when prompted to substantiate to construct the brand new Conda surroundings (
medical-image-ai).
Constructing the Conda surroundings will take as much as 5 minutes. - Within the terminal that opened within the step earlier than, run the next command to put in NodeJS within the
studiolabConda surroundings, which is required to put in the ImJoy JupyterLab 3 extension subsequent:conda set up -y -c conda-forge nodejs
We now set up the ImJoy Jupyter extension utilizing the Studio Lab Extension Supervisor to allow interactive visualizations. The Imjoy extension permits itkWidgets and different data-intensive processes to speak with native and distant Jupyter environments, together with Jupyter notebooks, JupyterLab, Studio Lab, and so forth. - Within the Extension Supervisor, seek for “imjoy” and select Set up.
- Verify to rebuild the kernel when prompted.
- Select Save and Reload when the construct is full.
After the set up of the ImJoy extension, it is possible for you to to see the ImJoy icon within the high menu of your notebooks.
To confirm this, navigate to the file browser, select the TCIA_Image_Visualalization_with_itkWidgets pocket book, and select the medical-image-ai kernel to run it.
The ImJoy icon will likely be seen within the higher left nook of the pocket book menu.
With these set up steps, you will have efficiently put in the medical-image-ai Python kernel and the ImJoy extension because the prerequisite to run the TCIA notebooks along with itkWidgets on Studio Lab.
Check the answer
We’ve created a set of notebooks and a tutorial that showcases the combination of those AI applied sciences in Studio Lab. Make sure that to decide on the medical-image-ai Python kernel when operating the TCIA notebooks in Studio Lab.
The first SageMaker notebook exhibits easy methods to obtain DICOM photographs from TCIA and visualize these photographs utilizing the cinematic quantity rendering capabilities of itkWidgets.
The second notebook exhibits how the professional annotations which can be accessible for a whole bunch of research on TCIA might be downloaded as DICOM SEG and RTSTRUCT objects, visualized in 3D or as overlays on 2D slices, and used for coaching and analysis of deep studying methods.
The third notebook exhibits how pre-trained MONAI deep studying fashions accessible on MONAI’s Mannequin Zoo might be downloaded and used to phase TCIA (or your personal) DICOM prostate MRI volumes.
Select Open Studio Lab in these and different JupyterLab notebooks to launch these notebooks within the freely accessible Studio Lab surroundings.
Clear up
After you will have adopted the set up steps on this put up and created the medical-image-ai Conda surroundings, it’s possible you’ll need to delete it to avoid wasting space for storing. To take action, use the next command:
conda take away --name medical-image-ai --allIt’s also possible to uninstall the ImJoy extension through the Extension Supervisor. Bear in mind that you’ll want to recreate the Conda surroundings and reinstall the ImJoy extension if you wish to proceed working with the TCIA notebooks in your Studio Lab account later.
Shut your tab and don’t neglect to decide on Cease Runtime on the Studio Lab venture web page.
Conclusion
SageMaker Studio Lab is accessible to medical picture AI analysis communities for free of charge and can be utilized for medical picture AI modeling and interactive medical picture visualization together with MONAI and itkWidgets. You should use the TCIA open information and pattern notebooks with Studio Lab at coaching occasions, like hackathons and workshops. With this resolution, scientists and researchers can shortly experiment, collaborate, and innovate with medical picture AI. When you’ve got an AWS account and have arrange a SageMaker Studio area, you can too run these notebooks on Studio utilizing the default Knowledge Science Python kernel (with the ImJoy-jupyter-extension put in) whereas choosing from a wide range of compute occasion sorts.
Studio Lab additionally launched a brand new function at AWS re:Invent 2022 to take the notebooks developed in Studio Lab and run them as batch jobs on a recurring schedule in your AWS accounts. Subsequently, you possibly can scale your ML experiments past the free compute limitations of Studio Lab and use extra highly effective compute situations with a lot larger datasets in your AWS accounts.
Should you’re considering studying extra about how AWS will help your healthcare or life sciences group, please contact an AWS representative. For extra info on MONAI and itkWidgets, please contact Kitware. New information is being added to TCIA on an ongoing foundation, and your ideas and contributions are welcome by visiting the TCIA website.
Additional studying
In regards to the Authors
Stephen Aylward is Senior Director of Strategic Initiatives at Kitware, an Adjunct Professor of Pc at The College of North Carolina at Chapel Hill, and a fellow of the MICCAI Society. Dr. Aylward based Kitware’s workplace in North Carolina, has been a pacesetter of a number of open-source initiatives, and is now Chair of the MONAI advisory board.
Matt McCormick, PhD, is a Distinguished Engineer at Kitware, the place he leads growth of the Perception Toolkit (ITK), a scientific picture evaluation toolkit. He has been a principal investigator and a co-investigator of a number of analysis grants from the Nationwide Institutes of Well being (NIH), led engagements with United States nationwide laboratories, and led numerous industrial initiatives offering superior software program for medical units. Dr. McCormick is a powerful advocate for community-driven open-source software program, open science, and reproducible analysis.
Brianna Main is a Analysis and Improvement Engineer at Kitware with a ardour for growing open supply software program and instruments that can profit the medical and scientific communities.
Justin Kirby is a Technical Undertaking Supervisor on the Frederick Nationwide Laboratory for Most cancers Analysis (FNLCR). His work is targeted on strategies to allow information sharing whereas preserving affected person privateness to enhance reproducibility and transparency in most cancers imaging analysis. His group based The Most cancers Imaging Archive (TCIA) in 2010, which the analysis neighborhood has leveraged to publish over 200 datasets associated to manuscripts, grants, problem competitions, and main NCI analysis initiatives. These datasets have been mentioned in over 1,500 peer reviewed publications.
Gang Fu is a Healthcare Resolution Architect at AWS. He holds a PhD in Pharmaceutical Science from the College of Mississippi and has over ten years of know-how and biomedical analysis expertise. He’s keen about know-how and the impression it might make on healthcare.
Alex Lemm is a Enterprise Improvement Supervisor for Medical Imaging at AWS. Alex defines and executes go-to-market methods with imaging companions and drives options growth to speed up AI/ML-based medical imaging analysis within the cloud. He’s keen about integrating open supply ML frameworks with the AWS AI/ML stack.
