CENTER-TBI

CENTER-TBI is a European observational study of traumatic brain injury, funded by the EU 7th Framework Programme and run by a consortium coordinated from Antwerp University Hospital. It set out to improve how TBI is characterised and to compare the effectiveness of different care practices across centres. Its core study followed patients of every severity who arrived within 24 hours of injury and needed a CT scan, and stored their images, blood samples and outcomes in central repositories for later research.

Composition

The core study enrolled 4,559 patients between December 2014 and December 2017, and 4,509 were analysed in the 2019 descriptive paper. The project website describes the database as more than 4,500 patients from 65 centres. Patients fall into three care pathways: discharged from the emergency room (848), admitted to a ward (1,523) and admitted to intensive care (2,138). Median age is 50 years, 28% are older than 65, and 67% are male. Outcomes were assessed at 6 months and collected up to 2 years after injury. A separate registry of about 22,800 patients holds administrative data only.

Acquisition

Every patient had a CT scan at presentation, with follow-up CT as clinically indicated. MRI was done in a subset of centres, and the 2019 paper analysed the first 504 early MRI studies (within 3 weeks of injury). The harmonised MRI protocol, mandatory for every MRI patient, has a 3D T1-weighted scan, T2-weighted, FLAIR, DTI and a T2* sequence (gradient echo and/or SWI), with optional resting-state fMRI. Protocol sheets exist for GE, Philips and Siemens scanners. Images were defaced on upload and are hosted by icometrix.

Annotations

All uploaded images were read centrally with structured reporting based on the NINDS common data elements, alongside local reads recorded in the case report form. Reads cover findings such as contusions, subdural and epidural haematoma, subarachnoid haemorrhage and traumatic axonal injury. There are no published voxel-level lesion masks.

Known limitations

• Access requires an approved study plan, a signed data use agreement and Management Committee review of manuscripts; commercial products and redistribution are not allowed.
• MRI covers only part of the cohort, and the total number of MRI patients is not published.
• Centres are mostly referral centres for neurotrauma and children are under-represented.
• Dataset versions differ (Neurobot, then Opal from 2022), so counts can change between versions.