Dear all,
It is my pleasure to officially announce the release of a new plugin that brings deep learning to the Orthanc ecosystem, in the context of breast imaging.
This plugin turns Orthanc into an integrated and interoperable platform for detecting breast masses on scanned mammograms. The plugin contains a state-of-the-art deep learning model (more precisely, a RetinaNet architecture) that is trained on the well-known CBIS-DDMS database.
The objective is certainly not to provide a clinical-ready solution, but rather to introduce physicians and skilled workers to AI for medical imaging using an entirely open ecosystem that they can directly run on their own personal computer. Another objective consists in popularizing the recent AI Results profile from IHE that prescribes the encoding of AI outputs as DICOM-SR (structured reports) using the 1500 and 1410 templates. The DICOM-SR instances are then rendered using a development version of the Stone Web viewer.
Interested persons are invited to read the reference article presented this week at the Medical Informatics Europe (MIE 2024) conference.
Kind Regards,
Sébastien-
