| Zugriffsnummer | 50008 |
| Dokumenttyp | Konferenzartikel |
| Peer Review | unbekannt |
| Sprache | Englisch |
| Titel | ScanHub: open-source platform for MR scanner control, acquisitions and postprocessing |
| Autor(in); Institution |
Schote, David; 8.1, Biomedizinische Magnetresonanz, PTB-Berlin; Brain-Link UG Landau i.d. Pfalz, GERMANY
Behrens, ; Brain-Link UG Landau i.d. Pfalz, GERMANY
Dinh, Ch.; Brain-Link UG Landau i.d. Pfalz, GERMANY
|
| Quelle/Jahr | Proceedings of the International Society for Magnetic Resonance in Medicine: 31 (2023), 2 S. |
| Artikelnummer | 2391 |
| Availability | [online only] |
| ISSN | 1545-4428 |
| DOI | |
| Konferenzangaben | 32th ISMRM & ISMRT Annual Meeting & Exhibition, Toronto, 03-08, June, 2023, Canada |
| Freie Schlagworte | software tools |
| Zusammenfassung | ScanHub (https://github.com/brain-link/scanhub-ui) is an open, generic solution for cloud-based medical data acquisition and processing. Functionalities are subdivided into microservices supporting use cases in the clinical as well as in the research context. The platform capabilities are demonstrated with an exemplary MRI workflow. As a proof of principle MR data acquisition was simulated with an open-source Bloch solver. A reconstruction of the simulated raw data is provided by a microservice. The whole acquisition process is controlled via a web-based UI; from deploying a pulse sequence to the organization and visualization of reconstructed and DICOMized results. Introduction A variety of open-source software in the field of MR examinations has already been published, addressing specific subproblems in the image acquisition and processing chain1, 2. These developments include software for: hardware-independent pulse sequence development3-5, MR simulations6, 7, image reconstructions8-10, data formats11, data viewers12, and image post-processing13, 14. ScanHub is a cloud-based open-source software that aims to merge and extend these available open-source software tools to operate open- and closed-source MRI scanners. The current functionality includes: selection of pulse sequences and MR protocols, display of pulse sequence parameters (RF and gradient pulses), performing image reconstructions as well as viewing and exporting data. Methods Event-Driven Microservice Architecture ScanHubs components are containerized17, structuring the platform in frontend, devices, and microservices. The different containers can run on different devices, whereby the setup can be suited to the individual requirements of the clinic or research site. ScanHubs component interfaces are largely built on the fast and performant web framework FastAPI18. By different HTTP routes, the React-based frontend communicates to the API. ScanHubs functionalities are subdivided into microservices. For example, a service for image reconstruction and a service for acquisition control. The division into individual services enables extensibility and scalability, i.e., adaptation to new use cases and execution on remote servers, e.g., cloud resources, see figure 1. This separation also allows easier clearance of certain modules for future regulatory approval whereas others can remain experimental. Each microservice is running as a separate container, allowing the use of different programming languages. The microservices are loosely coupled via a message broker including an event store, fostering extensibility and maintainability. The message broker, which is realized by the open-source Kafka framework receives published messages, which are distributed via topics. According to the respective topic, the message broker sends messages to its subscribers, following the Publish/Subscribe messaging pattern19. Through this mechanism, microservices can be triggered by topic-related events. This message structure enables the organization of various dynamic processing workflows. The browser user interface and the connected MR device communicate with the microservices via the API gateway. Acquisition and processing data are transferred to a data storage. MR Workflow To organize MR acquisitions on the platform, metadata of devices, patients, procedures, and records is stored in a data storage, here a PostgreSQL20 database. The user interacts with the platform through a browser-based react UI. Devices register themselves at the respective device control microservice via an API. It creates dedicated topics for acquisition control and monitoring within the message broker following the already mentioned Publish/Subscribe pattern. To perform an MR acquisition a sequence in the open-source PulSeq framework must be selected or uploaded. When the acquisition is executed, the sequence is communicated to the device. This allows a flexible hardware-independent approach with the interpreter of the PulSeq sequence running on the connected device. For now, the device is a virtual MRI machine that uses an open-source Bloch solver implemented in Python and Julia for simulations 21. The reconstruction of the simulated raw data in DICOM format is provided by a microservice and addressed through the respective topic. It can be run on a dedicated server with sufficient performance. For now, data is reconstructed by FFT, but in general, this can be replaced by any reconstruction framework. The simulated raw and reconstructed DICOM data are stored in an Orthanc22 database. The reconstructed image is visualized in the ScanHub UI enabled by an integrated web-based DICOM viewer 23. |
| Themenbereich der Metrologie | Metrologie in der Medizin |
Zitierung
Schote, D., Behrens, Winter, L., Kolbitsch, C., & Dinh, C. (2023). ScanHub: open-source platform for MR scanner control, acquisitions and postprocessing. 32th ISMRM & ISMRT Annual Meeting & Exhibition, Toronto, 03-08, June, 2023, Canada. https://doi.org/10.58530/2023/2391