FHIR has quickly emerged as a crucial standard for the interoperability of clinical data. While FHIR supports medical imaging through its ImagingStudy resource, the absence of openly available software combining FHIR with DICOM limits the development of applications linking clinical and imaging data. On another front, deep learning applied to medical imaging is becoming increasingly popular, and it is essential to educate a broad audience about its possibilities. However, the current lack of open, easy-to-use environments capable of executing a library of open-access models directly on DICOM images and on a standard computer poses an issue from a pedagogical perspective. In this paper, the Orthanc server is presented as a promising solution to both challenges. Plugins are developed to enable Orthanc to call software libraries written in the Java and Python programming languages. These plugins are then used to turn Orthanc into a FHIR server using the HAPI framework, as well as into a platform for running the inference of a deep learning model for breast imaging using either PyTorch or Deep Java Library. The resulting source code is released as free and open-source software, aiming to promote support for medical imaging in FHIR, to share technological knowledge about medical interoperability and deep learning, and to provide a test environment for the integration of imaging data into clinical workflows.