Since 1999, the Insight Toolkit (ITK) has been revolutionizing the way medical images are explored and analyzed. In addition to providing a robust infrastructure for the dissemination of complex algorithms across different fields, ITK has organically created a vibrant and active community in scientific image analysis.
In an attempt to objectively measure the impact of this open-source tool and the number of its active users, last year we launched a community survey to hear first hand about the ITK community. We received a total of 82 responses. Below we summarize a few of the findings.
About the ITK Users
The majority of users of ITK work in academic research (66%) followed closely by those working in commercial software development (32%). The overwhelming majority of ITK users work in the field of medical imaging (74.4%). We find ITK is also used in other fields such as biological imaging (12%), material science (3.7%) or biomechanics (1.2%).
Our survey revealed the fact ITK’s C++ programming interface is most often used, followed closely by ITK Python programming interface.
When asked about how ITK is used, we obtained responses in three main categories: a) Direct use, b) use as part of derived products that heavily rely in ITK such as ANTS, RTK, 3D Slicer or MeVisLab and c) as a resource in their own image analysis codebases, examples of this are QUIT, qMRLab, or Allen Institute’s aics-segmentation. The amount of possible scientific research enabled by ITK is illustrated by the hundreds of publications that were referenced in the survey.
ITK into the Future and Beyond
Our users also pointed out areas of improvement they would like to see in our toolbox including GPU acceleration, needed for the processing of larger datasets, as well as the improved interfaces with deep learning libraries. These consistent with are the standard trends in the image processing field.
Figure generated using wordclouds.com from the words found in the “What areas of improvement or features would you like to see in future versions?” and “How could ITK better serve your research and product development needs?” questions of the survey. 56% of the users that responded to the survey provided this feedback.
It is clear that ITK ought to follow current trends in technology; GPU accelerated processing for large data and machine learning. It is also clear that despite having plenty of processing filters for segmentation and registration, ITK also should continue making ensure adoption of new method to make the state of the art in those domains available for general adoption.
When asked about how to strengthen the community, the opinion was expressed to increase the amount of examples and improve the style of the existing examples to adhere to modern style guidelines. Also, improve the quantity of tutorials on how to perform basic image operations with ITK classes, e.g. webinars and video tutorials or improve and update the ITK Guide. Finally, more hackathons in which the community can exchange ideas in person are desired.