Software Tools for DICOM Media Exchange in Clinical Research: the SWABIK Project Software Tools for DICOM Media Exchange in Clinical Research: the SWABIK Project O. ElGazzar 1, M. Onken 1, M. Eichelberg 1, A. Hein 2 1 OFFIS e.v, Healthcare R&D, Oldenburg, Germany 2 Carl von Ossietzky-Universität, Dept. of Computer Science, Oldenburg, Germany This work is funded by the German federal ministry of research and education (grant 01 EZ 1023)
2 Outline Introduction Clinical Research Workflow: A Use Case Aim of the SWABIK Image Selection Image Pseudonymization Media Export Media Import Quality Assurance Manual Automated Conclusion
3 Introduction Medical imaging plays an increasingly important role in extracting quantitative measurements for clinical trials. Evaluating the efficacy of a chemotherapy drug Assessing a method of treatment for radiotherapy in cancer trials. DICOM is used for the exchange of images for clinical use and research purposes. Clinical routine: For patient care Clinical research: For interchange of images in clinical trials DICOM storage media (so called patient CD/DVD) are usually used for the exchange of images in clinical trials.
4 Clinical Research Workflow: A Use Case 1) A pharmaceutical company (i.e sponsor) wants to evaluate the efficacy of a chemotherapy drug. 2) The sponsor specifies the protocol that describes the objective(s), design, methodology, statistical considerations, and organization of a trial. 3) The sponsor asks imaging sites to participate in the trial. 4) The sites evaluate the RFP of the sponsor and the clinical appropriateness of the trial to the site capabilities and reports it to the sponsor. 5) The sponsor selects the appropriate site(s) and the site is identified by an identifier. 6) The principal site investigator (i.e. physician) receives the protocol from the sponsor.
5 Clinical Research Workflow: A Use Case (cont.) 7) A patient who meets the selection criteria of the clinical trial protocol is selected by a physician to be enrolled in a certain clinical trial. A physician s practice would be clinical research along with patient treatment. 8) The patient signs his consent approval to share his medical images. 9) The patient signs his consent approval to share his medical images. 10) A unique identifier for the patient as a subject is generated by the coordination center. 11) Patient s images are acquired, collected, pseudonymized and sent on a CD/DVD to the center. 12) In the center, images are blindly reviewed and evaluated by radiologists. 13) The clinical trial analysis results are reported to the sponsor.
6 Clinical Research Workflow: A Use Case
7 Aim of the SWABIK Project Supporting exchange of DICOM Storage Media (CD/DVD) in clinical research. To improve the efficiency and effectiveness of the clinical research workflow. Best-practice guidelines for supporting users who are creating and importing DICOM media in the clinical trials context. Development of standardized open source software tools including A tool for image selection A tool for image pseudonymization A tool for quality assurance A tool for media export A tool for media import A tool for recording tumor measurements (based on RECIST criteria) The project is funded by the German federal ministry of research and education (BMBF) (grant 01 EZ 1023). http://www.dicom-cd.org/ The project started on 01.09.2010 and it is running for 2 years!
8 Image Selection While a user is viewing images, he can select a study or series of interest for a subject in a clinical trial. A key object selection document (manifest) is generated to identify the user s selection. [IHE TCE Profile] Field Title Description Text Observer Observation Date Time Delay Reason Content Always automatically set to For Clinical Trial Export Clinical trial protocol identifier Person who selected this assembly of images Date and time when images have been selected Required if the export to be delayed until a certain reason is satisfied
9 Image Pseudonymization Patient identification information has to be removed to protect the patient s privacy. However, There could some attributes required for conducting a meaningful clinical trial analysis (e.g. Patient s Weight, Patient s Age,..etc). DICOM supplement 142: Clinical Trial De-identification Profiles was developed by DICOM WG 18 to specify a de-identification strategy by using confidentiality profiles for common DICOM attributes in clinical trials. Help implementors to be compliant Help sites to configure their own profile based on the trial requirements If re-identification of some patient attributes is required, these attributes are encrypted and stored in the Encrypted Sequence DICOM attribute. Those attributes could be restored back to the DICOM dataset by decrypting the encrypted sequence using the appropriate key. For data Integrity and authentication Use a digital signature
10 DICOM Media Export DICOM media are problematic! CDs are not readable Bad image quality The problem of handling DICOM media in clinical routine was addressed by the DRG project www.dicom-cd.de However, not ín the clinical research context! Media of high quality are important for a successful workflow in clinical trials IHE PDI profile and DRG specifications are highly recommended for media creators Already available DICOM validation tools are not appropriate for clinical research Show hundreds of errors in a format not understandable for non-technical staff! A new concept for media quality assurance is required
11 DICOM Media Import The media import workflow is not clear in clinical research New guidelines are required to adapt to the case of clinical trials in order to Consider meta-data attached to the medical images Case Report Form (CRF) Data Transmittal Form (DTF) Consider privacy of patient identification information Consider medical image quality assurance Media should be checked before being imported Development of a tool for QA is required A simplified concept for media QA is needed
12 Quality Assurance Definition. Quality Assurance (QA): All those planned and systematic actions that are established to ensure that the trial is performed and the data are generated, documented (recorded), and reported in compliance with good clinical practice (GCP) and the applicable regulatory requirement(s). It is recommended to perform QA procedures twice in the workflow Before media export Before media import Quality assurance is important for a consistent and reproducible image analysis Ensure consistency between subjects Ensure consistency for a subject at different time points A dedicated workstation is often used for performing Quality Assurance checks Manual visual inspection (qualitative) Automated check (quantitative)
13 Quality Assurance: Visual Inspection Check that patient s privacy is protected as specified by the protocol. Check for the coverage of the anatomy of interest (e.g. chest, abdomen, pelvis). Check for the patient positioning during the scan. Check for the contiguous patient scanning (i.e. no slices are missed). Check for the image Field of View (FOV). Check for specific attribute values required by the clinical trial protocol. (e.g. KVp, mas) Check for the absence of image artifacts. (e.g. external or internal patient motion due to breathing during the scan could lead to undesired motion artifacts in the image). Check for the overall quality of images. (e.g. inappropriate lossy image compression).
14 Quality Assurance: Automatic Check Validation of the encoding of the DICOM header attributes. Simplified format for the validation results (e.g. yes/no signal) Show error details if required Configured for application-specific attributes (e.g. MPR) Configured for protocol-specific attributes (e.g. CT) Example: CT protocol-specific attributes Technical Parameter kvp X-ray tube current Exposure Time Exposure in mas Reconstruction Interval Reconstruction Thickness Reconstruction Filter Reconstruction Overlap Pitch Description Peak kilo voltage of the output x-ray generator X-ray tube current in ma Time of X-ray exposure in ms The exposure expressed in mas The spacing between the slices Nominal slice thickness, in mm The reconstruction algorithm The overlap between slices Ratio of the table feed rotation to the slice thickness
15 Software Requirements Open source Freely available for download Simple to use (also for doctors or other hospital staff) User manual Flexibility Should be configurable to support different types of clinical trials Portability Cross-platform (Windows, Linux, Mac OS X,...etc) Based on DCMTK and QT No other significant external dependencies (.Net framework, Java, ) Maintainability Extensible in the future Decent source code documentation
16 Conclusion Medical imaging is increasingly used for extracting quantitative measurements in clinical trials. A concept for a framework for quality assurance in clinical trials is required. Standardized software tools and guidelines are developed for supporting users exchanging medical image in the clinical trial contexts. Improve efficiency and effectiveness of clinical research workflow