VoxLogicA is a new system for analysing medical images in a declarative way. VoxLogicA is designed for domain experts, not for programmers. No libraries to download, no project files to create, no compilers to run, just describe your tasks using short text files, and run them in seconds!
VoxLogicA is still in a early stage of development. There are a bunch of interesting features, but something very important for your use case may still be missing. Right now, it is very important for us to create a community. We need use cases, reports, suggestions, collaboration, funding, and feedback of any type.
If interested, please, get in touch.
Aimed at non-programmers.
Easy to share, easy to check.
No entry fee, open to contribution.
When you use VoxLogicA, your source code
typically is in the order of 10-30 lines of code (yes, we didn't add
any "K" after the numbers).
The language of VoxLogicA is close to the
domain of imaging, featuring constructs like
We have tested our approach on the BraTS dataset for glioblastoma segmentation. Our method is described in 10 lines of text and runs in seconds on a standard desktop. Its accuracy is similar to that of humans, and of best-in-class machine learning algorithms.Got curious? Read our TACAS 2019 publication
Inside VoxLogicA, there's a powerful computational engine called a Model Checker. VoxLogicA computes your analysis in parallel using all the available CPU cores. The result is implicitly backed up by a formal mathematical interpretation, guaranteed to yield the same result on Windows, Linux, OSX.
It does not take a lot to learn how to use VoxLogicA. Analysis is so short that you can share it to your colleagues, even in a chat window.
We need you, users, developers, and domain experts, to shape the future of our technology. Please get in touch if you are interested.