Overview
Slicebox is a microservice for safe sharing and easy access to medical images. The goal of the project is threefold:
- to facilitate research and collaboration between hospitals and universities,
- to act as a DICOM image server for group access to data, thus avoiding duplication,
- to provide a storage and communications platform for creators of medical imaging applications.
The service makes it easy to send images from the hosptial's PACS (image archive) to collaborators on the outside. Anonymization is handled automatically, releaving the hospital staff of the burden of making sure patient information does not leave the hospital network.
Slicebox provides a rich set of features for importing, exporting, managing and sharing image data. All features listed below are available as REST calls. The provided web interface covers most features.
- Browsing of images, either according to a Patient-Study-Series hierarchy, or using a flat view of all available series.
- Querying, filtering and sorting of image (meta) data
- Upload and download of image data (DICOM datasets)
- Viewing of images
- Listing of DICOM attributes
- Tagging series with arbitrary tags and filtering search results on one or more tags
- Mapping sets of key-value pairs of DICOM attributes to user defined series types for easy integration with applications
- Managing connections to other slicebox instances
- Sending images to and receiving images from other slicebox instances, with automatic anonymization of datasets
- Listing and searching anonymization information to map anonymized patient information to real patient data
- Setup of DICOM Service Class Providers (SCP). An SCP is a server accepting transfers of DICOM images. This functionality makes it possible to send images from a PACS (image archive) to Slicebox.
- Setup of DICOM Service Class Users (SCU). This is the opposite of an SCP. An SCU client makes it possible to export DICOM encapsulated images and results to PACS.
- Setup directory watches. DICOM files dropped into a directory watched by slicebox will import the image.
- Protection of resources and routes using authentication and authorization. Users with certain privileges can be added and removed.
Our canonical use case is a hospital research group collaborating with researchers in medical image analysis at a university. Slicebox instances are installed at both sites and are setup to communicate with each other. Furthermore, the hospital instance is setup to receive images from the hospital PACS system as shown in the image below.
Images can now be exported from PACS to the local slicebox instance easily. Exported images can be browsed and handled using slicebox's intuitive web interface. From here, images can be sent to the slicebox instance installed at the university. Slicebox will handle anonymization according to the DICOM standard before images leave the hosptial. On the university side, researchers can now access these images both using the web interface as well as an extensive REST api. Benefits of storing image data in slicebox as opposed to local files include
- All research group members have access to all data
- Code can be shared more easily since data paths will work across computers
- Organizing DICOM files is cumbersome. An API which implements the DICOM hierarchy of patient-study-series-image reduces this burden and reduces the risk of analyzing the wrong data
A second use case is the development and deployment of medical imaging applications. Application makers should focus on the functionality of app itself and not have to re-invent a system for integration in the clinical workflow. Using slicebox as an image data storage and communications platform for applications provides a general and configurable solution for local and/or cloud deployment of applications. Since slicebox can be managed using a REST programming API, most software projects can be easily integrated with slicebox.