Back to Projects List
Deep Learning Integration in Slicer
- Jorge Onieva (BWH)
- Raúl San José (BWH)
- Roya Khajavi
- Alireza Mehrtash
- Andrey Fedorov
Integrate a lung segmentation algorithm based on Deep Learning in Slicer.
- Develop a proof of concept to test the integration of Keras+Tensorflow tools in Slicer
- Create a Slicer package that can be distributed with these features
Approach and Plan
- Integrate the algorithm+pretrained models in CIP (see CIPDeepLearningLungSegmentation).
- Compile Slicer against a customized Python that includes all the CIP required components
- Pack Slicer with that Python version
Progress and Next Steps
- This integration was done through the CustomSlicerGenerator in MacOS and Linux.
- Luckily, it would be obsolete in Slicer 5!! A template with a Python distribution based on Anaconda or others may be used
- Also, we found out other extensions like DeepInfer and TOOMCAT that may be useful in the meantime