This subproject is one of many research subprojects utilizing theresources provided by a Center grant funded by NIH/NCRR. The subproject andinvestigator (PI) may have received primary funding from another NIH source,and thus could be represented in other CRISP entries. The institution listed isfor the Center, which is not necessarily the institution for the investigator.Functional magnetic resonance imaging (fMRI) is a specialized application of MRI which places additional demands on scanner performance and stability compared to routine clinical studies. Since fMRI attempts to identify brain regions activated in the presence of particular stimuli or tasks, instabilities or excessive noise in the MRI signal can lead to either false positive or negative detection of activation. When applied to group comparisons as is commonly the case in the study of disease or pharmacological interventions, changes in data quality over time can be misinterpreted as differences between groups. The requirements for fMRI generally exceed the specifications covered by commercial manufacturer quality assessment (QA) and preventative maintenance.To address these needs, we have developed QA methods tailored for fMRI data. Specifically, several metrics are computed from an fMRI data set to test for temporal stability, overall signal-to-noise ratio (SNR), and scanner signal spiking , and subject motion. Moreover, software was developed to cull the data storage archives of the Resource. Using a completely automated system, the QA metrics were assessed on over 1000 fMRI studies that have been conducted since 2004. This task would be insurmountable without complete automation of data detection and analysis.The results reveal that periods of scanner instability are observed, despite regular maintenance by the commercial provider. In addition, differences in data quality due to experimental protocol, such as selection of imaging coil, are readily apparent. With this information, safeguards against over-interpretation of substandard data can be implemented. Retrospective assessment of the results will also be used to support optimization of experimental protocols.
Showing the most recent 10 out of 414 publications
