Back to Projects List
Chest Imaging Platform: segmentation of lungs and pectoralis based on deep learning
- Jorge Onieva (BWH)
- Raúl San José (BWH)
- Pietro Nardelli (BWH)
Integration of a deep learning infrastructure that supports future integration of deep learning - based tools.
Two use cases: lung segmentation and pectoralise segmentation
- Integrate all the needed infrastructure components and required external libraries: Keras + tensorflow, etc.
- Creation of CLIs as a use case
Progress and Next Steps
- The integration of Keras and Tensorflow had some issues in some platforms.
- Nevertheless, we are fixing the last issues and it should be ready next week.
- We are also working in speeding up the lung segmentation algorithm in order that CPU running is feasible
Background and References
The Chest Imaging Platform is an open source project that can run as an standalone and it is integrated in Slicer as part of the Slicer solutions. More info in https://chestimagingplatform.org
- Source code: https://chestimagingplatform.org/
- Documentation: https://chestimagingplatform.org/documentation
- Test data: http://cdash.chestimagingplatform.org/index.php?project=CIP