1. Research database structure
● The original (DICOM) imaging data acquired by MRI and CT methods are provided with conversion to NIfTI format. Secondary data including cerebrovascular segmentation and skeletonization are additionally available as NIfTI format. The conversion procedures from original source to NifTI format is available by NITRC-provided dcm2nii framework.
● Algorithm-assisted multi-dimensional quantitative features and clinical characteristics are available by CSV format anonymized under personal information protection act.
● When employing conventional data, data-wise relationship, data type, collection, and forms of usage can be commensurate by the primary section as the host holder.
● When collecting multi-institutional data, the reliability is validated by committee-based evaluation composed of board certified neurologists and radiologists with outpatient treatment experience (3, 5, and 10 yrs).
● Data collection progresses following IRB authorization approved by Samsung Medical Center IRB board. Multi-institutional data collection proceeds conferred with profesionally involved research scientists.
2. Standards of research data and metadata
| Specification | TOF | FLAIR | DWI | ADC | GRE |
|---|---|---|---|---|---|
| Phase encoding | ROW | ROW | COL | COL | ROW |
| Field of View phase (%) | 80 | 76.25 | 100 | 100 | 76.56 |
| Slice thickness (mm) | 1.2 | 5 | 5 | 5 | 5 |
| Repetition time (ms) | 25 | 11000 | 3000 | 3000 | 676.85 |
| Echo time (ms) | 3.454 | 125 | 81.06 | 81.06 | 16.11 |
| Flip angle (degrees) | 20 | 90 | 90 | 90 | 18 |
| Bandwitdth (Hz/Px) | 124 | 221 | 2064 | 2064 | 217 |
3. Sharing and constraints of research data
● Research data sharing necessitates high performance computing server for large storage and data processing. Stream of data and network facilities interfaced with web database are necessary.
● User friendly web interface and domain purchase facilitate accesibility. Web server can be built outsourcing external computer network and security experts.
● The primary and secondary publicization timeline of research database are 2nd and 5th year respectively.
● Respite of unconcealing of database maintains until 2nd year interval of research project.
● The host to legally hold intellectual property is commensurate following the internal legal binding of the host institution (Samsung Medical Center) research support department. The legal boundary of sharing and utilizing an intellectual property is maximized to the ceiling of internal regulation.
4. Reuse and distribution of research dataset
● The authorization of research database usage covers radiologists, neurologists, brain scientists, and medical imaging professionals dedicated to cerebrovascular field verifiable by institutional certification. We strongly recommend against enterpreneurs and interest-binded individuals to use the database.
● The reusing and reprocessing of database towards users are enabled by thorough monitoring of cutting-edge brain research trends and algorithm optimization. Explicit improvement agenda are transparently disclosed to users for stabilized research activities.
5. Data Storage
● Data management, processing, and storage are controllable and executable by high-performance computing, including the continuation of backup. When the analysis algorithm of the database is improved, the existing secondary processing data is newly created and then managed in the direction of deletion through consultation. The construction site is to be built inside the Samsung Medical Center.
● We plan to use a SATA-based high-performance processor for data transfer with an online database-URL interface streamlining a real-time information exchange network.
● Even after the completion of the research project, patient data from the host institution and partner institutions are continuously added to develop into an extensive database. Existing and continuously added data are considered to have sufficient storage value in prospective and retrospective studies.
● For the sake of database storage and usage, imaging and clinical data are converted into NifTI and CSV format under the standard brain imaging data structure (BIDS) policy.
● Metadata and documentation for data re-utilization are provided with clinical information in CSV format.
