In his earlier post, Ishwar Chandramouliswaran introduced the objectives of the NCIP Hub, an online resource for research and collaboration in cancer informatics. As a scientific repository, the NCIP Hub can store community-generated data, tools, and other resources. Members can upload tools, conduct analyses, and collaborate, giving researchers the opportunity to engage with and leverage each other's expertise. This summer, my role as a NCI summer intern included exploring its capabilities in testing and preparing training materials, contributing content, performing community outreach, as well as supporting lead users and early adopters of NCIP Hub.

The interview in this post highlights how Dr. Jayashree Kapalthy-Cramer, an informaticist, member of the Quantitative Imaging Network (QIN), and an Assistant Professor of Radiology at Harvard Medical School, has been using NCIP Hub to collaborate with her colleagues in the cancer imaging informatics community.

How have you been using the NCIP Hub?

The NCI's Quantitative Imaging Network has been piloting the use of NCIP Hub to meet certain imaging research community needs. It became apparent that the QIN needed a collaborative platform to support its long-term data and tool-sharing needs. This includes the ability to share data and image analysis tools, as well as statistical analysis algorithms, especially in the context of the current collaborative activities within the QIN working groups. To that end, we have been evaluating the suitability of NCIP Hub to support collaborative activities, including "challenges" designed to compare the performance of image analysis algorithms. Challenges require researchers to run their own algorithms on a shared data set, so a collaborative repository is essential.

What features of NCIP Hub have you been using?

We have established Groups' within NCIP Hub to collaborate with each other and have co-located tools and data for evaluation in shared Linux Workspaces" accessible via a web browser. We have also used Projects" to post results of experiments to share among the project team. In addition, the QIN regularly conducts meetings and workshops, the proceedings of which have been made available as Resources" in NCIP Hub.

What needs have you attempted to address, and how did you address them?

We have developed tools (R and MATLAB-based) in the workspace and are working on deploying these.

We also wanted to perform image visualization and analyses directly on NCIP Hub where the data and tools already reside. To do so, we needed to install the 3D Slicer tool, a web-based, open source software package. In general, NCIP Hub blocks network access to online tools unless whitelisted, or authorized for use. Currently, we have whitelisted the 3D Slicer site, as well as The Cancer Imaging Archive (TCIA) site. With access to these sites, we can now retrieve data from TCIA and perform analyses with the 3D Slicer. Installing the 3D Slicer extensions has been easy to do on NCIP Hub. However, this is a work-in-progress, and we are continuing to explore capabilities in NCIP Hub to more specifically address our needs.

How have you approached importing and exporting data?

There are several ways I have been uploading data to the NCIP Hub for analysis. I can import data directly from TCIA or I can upload local data to the NCIP Hub. The results of my analyses are directly saved into my workspace, and I can then export or share the results. We have started using NCIP Hub instead of Dropbox or Google Drive to share the output of our analyses.

What else would you like to share regarding your experience using the NCIP Hub?

NCIP Hub has a lot of potential, and we are continuing to experiment with using the platform to support the needs of QIN investigators. We have found it helpful to co-locate data, tools, workspaces, and shared code repositories. It"s also very useful to have the ability to share analysis sessions so that QIN collaborators can work in the same areas, and save these sessions and/or results to access them at a later time. We are still fleshing out the workflows associated with running tools on large, common datasets. Overall, the NCIP Hub environment offers a number of capabilities and utilities that address QIN community needs.