P2‐401: Reliability of DTI: A european multicenter study

Alzheimer's disease (AD) causes cortical neuronal loss with degenerative changes of white matter fiber tracts. Several studies showed the potential of diffusion tensor imaging (DTI) to detect early stages of white matter damage. For the development of future clinical markers we accessed the security and reliability of DTI in a multicenter setting. Using a standardised DTI phantom (developed by DKFZ, Heidelberg) we collected DTI data on 15 different MRI scanners at 11 clinical and radiological expert centers. In addition, we performed cranial MRI and DTI scans in a healthy young woman with the same scanners. The acquisition protocols for DTI were standardized as far as possible. The data were analyzed in respect to data quality and in respect to the variability of fractional anisotropy measures using manual region of interest measurement with DTI-Studio (Version 3.0.1) and tract based spatial statistics. Phantom data showed large differences in DTI scan quality: eight of 15 measurements had to be excluded due to scan artefacts related to field inhomogenities. From the remaining set of data with sufficient data quality, the mean phantom specific FA value across these scanners was 0.6, consistent with the degree of organization of the phantom fiber tract. The coefficient of variation across scanners ranged from 5.0% to 9.8%, depending on the localization of the sampling region. Inter-scanner variability of cerebral DTI data will be presented, as well. Our data indicate that multicenter DTI poses high demands on scanner maintenance and acquisition protocols. A phantom study is indispensable before a multicenter DTI diagnostic trial should be conducted. The phantom data give an upper estimate of between scanner variability of in vivo data, since the larger shimvolume and higher experience with in vivo DTI will likely reduce variation between scanners with brain data.