This page includes supplemental materials for the paper, FBIRN Recommendations for Multi-Center fMRI Studies:

The Function Bioinformatics Research Network (fBIRN) is a test bed within the national BIRN initiative. FBIRN was funded by the National Institute of Health (NIH) to develop data-sharing infrastructure, software tools, strategies and advisory services to make multi-site functional MRI studies a common research practice. Several large multi-site datasets consisting of fMRI, structural imaging, and clinical data were acquired during the FBIRN lifespan to develop and test these novel tools and methodologies. A total of 5 human data sets were acquired as part of fBIRN: the phase I traveling subjects data set, the phase II data set, the East coast travel subjects data set, the West coast traveling subjects data set, and the phase III data set.

The Phase I Traveling Subjects study was the first fBIRN multi-center study, and was acquired in 2003^1-5. Five healthy subjects were imaged on two occasions on 10 different scanners located in geographically diverse locations within the United States. The purpose of this study was to provide an excellent dataset with which to assess test-retest and between-site reliability of fMRI and to provide a rich data set to test tool and methods to allow for calibration of between site differences in FMRI results. Imaging included T1 wieghted scans, two cognitive fMRI tasks (auditory oddball (AO) and Serial Item Recognition Paradigm (SIRP) tasks), two calibration scans (breath hold (BH) and sensory motor (SM) tasks), and a resting state scan. Importantly, while identical tasks protocols were used across sites, no attempt was made to standardize some important scan parameters (scanner vendor, field strength, k-space trajectory, echo time, head coil type, etc), instead each site was allowed to use their “best” imaging protocol. Methods were developed to correct the data for the effects of site variability within the imaging data in a post hoc manner. Results of multiple analyses found^2-5 that while calibration methods and image analysis strategies were successful in removing much of the site variability, significant site dependence remained even after these methods were employed. The FBIRN collaboration has made the complete Phase I data set available to the scientific community for further methods development; these data can be downloaded from the fBIRN data repository website (http://fbirnbdr.nbirn.net:8080/BDR/).

The FBIRN Phase II study took place in 2005. Individuals with schizophrenia and healthy controls were imaged using structural and functional MRI imaging, and behavioral data, demographic information, and clinical assessments we also acquired on 253 subjects from around the US^6-8. Subjects were recruited locally, evaluated with a standardized clinical assessment battery, and scanned twice at each of the 9 participating FBIRN sites. A standardized protocol was used at each site which included the same cognitive (AO and SIRP) and calibration (SM and BH) fMRI tasks that were piloted during the fBIRN phase I traveling subjects study. As with the phase I study, while many scan parameters were matched across sites, important scan parameters were not the same across sites^6-8. Due to the large cohort of subjects imaged during phase II, analysis results from phase II data elucidated important differences between persons with schizophrenia and matched controls in the auditory oddball and working memory tasks as well in genetics and structural differences. The FBIRN collaboration has made the complete phase II data set to the scientific community for further methods development; these data can be downloaded from the BIRN data repository website (http://fbirnbdr.nbirn.net:8080/BDR/).

The third FBIRN study was the East Coast Traveling Subjects study (ECTS), which took place in 2007. Eighteen healthy participants were studied once at each of three magnet sites and twice at a fourth site. All sites were located on the East Coast of the United States. The purpose of this study was to provide a second dataset with which to assess test-retest and between-site reliability, to provide a rich data set to test tool and methods to allow for calibration of between site differences in FMRI results, and to evaluate how successful FBIRN site standardization methods were in reducing inter-site variability in fMRI data. The imaging protocol included a T1 weighted scan, seven runs of a working memory fMRI task, two calibration scans (breath hold (BH) and pulses arterial spin labeling (PASL)), and two resting state scans. Importantly, standardization methods that were identified and developed based on analysis from the previous two fBIRN studies were applied to this study. These included the use of a standard field strength (3T), us of vendor supplied receive only multi-channel head coils, and an EPI based k-space acquisition. In addition, a standard scanner check list was used during acquisition, scanner phantom QA data was acquired at all sites and metrics were actively monitored before and during data acquisition, human QA was performed during data acquisition, and a traveling engineer was sent to each site before the start of the study to provide final site evaluation and site standardization. Results from a multi-center analysis of the working memory data shows that between site variability is much smaller than between subject variability, suggesting that a dedicated effort of site standardization can generate a study with little residual inter-site variability ^{Brown 2011}. The FBIRN collaboration will make the complete ECTS data set available to the scientific community in 2011 on the BIRN data repository web site (http://fbirnbdr.nbirn.net:8080/BDR/).

The FBIRN West Coast traveling subjects study took place in 2008-9. Ten healthy participants and 10 participants with schizophrenia were studied twice at one site and once at each of two magnets at a second site, both located in Southern California. The purpose of this study was to provide a dataset s to evaluate test-retest and between-site reliability in both healthy participants and participants with schizophrenia. These data provide a rich data set to test tool and methods to allow for calibration of between site differences in FMRI results, and to evaluate how successful FBIRN site standardization methods were in reducing inter-site variability in fMRI data. The imaging protocol included the same imaging protocol as was used in the ECTS study. Methods developed by FBIRN to standardize multi-center fMRI studies were employed on each scanner. Analysis results from these multi-center data are ongoing and no results are available at this time. FBIRN collaboration will make the complete WCTS data set available to the scientific community in 2012 on the BIRN data repository web site (http://fbirnbdr.nbirn.net:8080/BDR/).

The final FBIRN study, phase III, collected data from 2010-11. Individuals with schizophrenia and healthy controls were imaged using a protocol including structural, perfusion, diffusion tensor, and fMRI scans. Behavioral data, clinical assessments, demographic data, and genetic data we acquired on all subjects. Subjects were recruited locally, evaluated with a standardized clinical assessment battery, and scanned once at each of the 7 participating FBIRN sites. The total recruitment goal was 210 participants with schizophrenia and 210 age and gender matched controls. The imaging protocol included a T1 weighted scan, seven runs of a working memory task, two runs of the phase II AO task, a PASL calibration scan, and resting state scans. As with the East and West coast traveling subjects studies, standardization of all aspects of the study was done to the extent possible. The large cohort of participants with schizophrenia and age and gender matched healthy are expected to elucidated additional important differences between subject populations and subpopulations within the cohorts in both the auditory oddball and working memory tasks, as well uncover important genetics and structural differences. Data analysis is currently underway. The FBIRN collaboration expects to make the complete phase III data set to the scientific community for further methods development as well as scientific investigations of schizophrenia; these data are expected to be made available in 2012 on the BRIN data repository website (http://fbirnbdr.nbirn.net:8080/BDR/).

This page includes supplemental materials for the paper, FBIRN Recommendations for Multi-Center fMRI Studies:

The Function Bioinformatics Research Network (fBIRN) is a test bed within the national BIRN initiative. FBIRN was funded by the National Institute of Health (NIH) to develop data-sharing infrastructure, software tools, strategies and advisory services to make multi-site functional MRI studies a common research practice. Several large multi-site datasets consisting of fMRI, structural imaging, and clinical data were acquired during the FBIRN lifespan to develop and test these novel tools and methodologies. A total of 5 human data sets were acquired as part of fBIRN: the phase I traveling subjects data set, the phase II data set, the East coast travel subjects data set, the West coast traveling subjects data set, and the phase III data set.

The Phase I Traveling Subjects study was the first fBIRN multi-center study, and was acquired in 2003^1-5. Five healthy subjects were imaged on two occasions on 10 different scanners located in geographically diverse locations within the United States. The purpose of this study was to provide an excellent dataset with which to assess test-retest and between-site reliability of fMRI and to provide a rich data set to test tool and methods to allow for calibration of between site differences in FMRI results. Imaging included T1 wieghted scans, two cognitive fMRI tasks (auditory oddball (AO) and Serial Item Recognition Paradigm (SIRP) tasks), two calibration scans (breath hold (BH) and sensory motor (SM) tasks), and a resting state scan. Importantly, while identical tasks protocols were used across sites, no attempt was made to standardize some important scan parameters (scanner vendor, field strength, k-space trajectory, echo time, head coil type, etc), instead each site was allowed to use their “best” imaging protocol. Methods were developed to correct the data for the effects of site variability within the imaging data in a post hoc manner. Results of multiple analyses found^2-5 that while calibration methods and image analysis strategies were successful in removing much of the site variability, significant site dependence remained even after these methods were employed. The FBIRN collaboration has made the complete Phase I data set available to the scientific community for further methods development; these data can be downloaded from the fBIRN data repository website (http://fbirnbdr.nbirn.net:8080/BDR/).

The FBIRN Phase II study took place in 2005. Individuals with schizophrenia and healthy controls were imaged using structural and functional MRI imaging, and behavioral data, demographic information, and clinical assessments we also acquired on 253 subjects from around the US^6-8. Subjects were recruited locally, evaluated with a standardized clinical assessment battery, and scanned twice at each of the 9 participating FBIRN sites. A standardized protocol was used at each site which included the same cognitive (AO and SIRP) and calibration (SM and BH) fMRI tasks that were piloted during the fBIRN phase I traveling subjects study. As with the phase I study, while many scan parameters were matched across sites, important scan parameters were not the same across sites^6-8. Due to the large cohort of subjects imaged during phase II, analysis results from phase II data elucidated important differences between persons with schizophrenia and matched controls in the auditory oddball and working memory tasks as well in genetics and structural differences. The FBIRN collaboration has made the complete phase II data set to the scientific community for further methods development; these data can be downloaded from the BIRN data repository website (http://fbirnbdr.nbirn.net:8080/BDR/).

The third FBIRN study was the East Coast Traveling Subjects study (ECTS), which took place in 2007. Eighteen healthy participants were studied once at each of three magnet sites and twice at a fourth site. All sites were located on the East Coast of the United States. The purpose of this study was to provide a second dataset with which to assess test-retest and between-site reliability, to provide a rich data set to test tool and methods to allow for calibration of between site differences in FMRI results, and to evaluate how successful FBIRN site standardization methods were in reducing inter-site variability in fMRI data. The imaging protocol included a T1 weighted scan, seven runs of a working memory fMRI task, two calibration scans (breath hold (BH) and pulses arterial spin labeling (PASL)), and two resting state scans. Importantly, standardization methods that were identified and developed based on analysis from the previous two fBIRN studies were applied to this study. These included the use of a standard field strength (3T), us of vendor supplied receive only multi-channel head coils, and an EPI based k-space acquisition. In addition, a standard scanner check list was used during acquisition, scanner phantom QA data was acquired at all sites and metrics were actively monitored before and during data acquisition, human QA was performed during data acquisition, and a traveling engineer was sent to each site before the start of the study to provide final site evaluation and site standardization. Results from a multi-center analysis of the working memory data shows that between site variability is much smaller than between subject variability, suggesting that a dedicated effort of site standardization can generate a study with little residual inter-site variability ^{Brown 2011}. The FBIRN collaboration will make the complete ECTS data set available to the scientific community in 2011 on the BIRN data repository web site (http://fbirnbdr.nbirn.net:8080/BDR/).

The FBIRN West Coast traveling subjects study took place in 2008-9. Ten healthy participants and 10 participants with schizophrenia were studied twice at one site and once at each of two magnets at a second site, both located in Southern California. The purpose of this study was to provide a dataset s to evaluate test-retest and between-site reliability in both healthy participants and participants with schizophrenia. These data provide a rich data set to test tool and methods to allow for calibration of between site differences in FMRI results, and to evaluate how successful FBIRN site standardization methods were in reducing inter-site variability in fMRI data. The imaging protocol included the same imaging protocol as was used in the ECTS study. Methods developed by FBIRN to standardize multi-center fMRI studies were employed on each scanner. Analysis results from these multi-center data are ongoing and no results are available at this time. FBIRN collaboration will make the complete WCTS data set available to the scientific community in 2012 on the BIRN data repository web site (http://fbirnbdr.nbirn.net:8080/BDR/).

The final FBIRN study, phase III, collected data from 2010-11. Individuals with schizophrenia and healthy controls were imaged using a protocol including structural, perfusion, diffusion tensor, and fMRI scans. Behavioral data, clinical assessments, demographic data, and genetic data we acquired on all subjects. Subjects were recruited locally, evaluated with a standardized clinical assessment battery, and scanned once at each of the 7 participating FBIRN sites. The total recruitment goal was 210 participants with schizophrenia and 210 age and gender matched controls. The imaging protocol included a T1 weighted scan, seven runs of a working memory task, two runs of the phase II AO task, a PASL calibration scan, and resting state scans. As with the East and West coast traveling subjects studies, standardization of all aspects of the study was done to the extent possible. The large cohort of participants with schizophrenia and age and gender matched healthy are expected to elucidated additional important differences between subject populations and subpopulations within the cohorts in both the auditory oddball and working memory tasks, as well uncover important genetics and structural differences. Data analysis is currently underway. The FBIRN collaboration expects to make the complete phase III data set to the scientific community for further methods development as well as scientific investigations of schizophrenia; these data are expected to be made available in 2012 on the BRIN data repository website (http://fbirnbdr.nbirn.net:8080/BDR/).

This page includes supplemental materials for the paper, FBIRN Recommendations for Multi-Center fMRI Studies:

The Function Bioinformatics Research Network (fBIRN) is a test bed within the national BIRN initiative. FBIRN was funded by the National Institute of Health (NIH) to develop data-sharing infrastructure, software tools, strategies and advisory services to make multi-site functional MRI studies a common research practice. Several large multi-site datasets consisting of fMRI, structural imaging, and clinical data were acquired during the FBIRN lifespan to develop and test these novel tools and methodologies. A total of 5 human data sets were acquired as part of fBIRN: the phase I traveling subjects data set, the phase II data set, the East coast travel subjects data set, the West coast traveling subjects data set, and the phase III data set.

The Phase I Traveling Subjects study was the first fBIRN multi-center study, and was acquired in 2003^1-5. Five healthy subjects were imaged on two occasions on 10 different scanners located in geographically diverse locations within the United States. The purpose of this study was to provide an excellent dataset with which to assess test-retest and between-site reliability of fMRI and to provide a rich data set to test tool and methods to allow for calibration of between site differences in FMRI results. Imaging included T1 wieghted scans, two cognitive fMRI tasks (auditory oddball (AO) and Serial Item Recognition Paradigm (SIRP) tasks), two calibration scans (breath hold (BH) and sensory motor (SM) tasks), and a resting state scan. Importantly, while identical tasks protocols were used across sites, no attempt was made to standardize some important scan parameters (scanner vendor, field strength, k-space trajectory, echo time, head coil type, etc), instead each site was allowed to use their “best” imaging protocol. Methods were developed to correct the data for the effects of site variability within the imaging data in a post hoc manner. Results of multiple analyses found^2-5 that while calibration methods and image analysis strategies were successful in removing much of the site variability, significant site dependence remained even after these methods were employed. The FBIRN collaboration has made the complete Phase I data set available to the scientific community for further methods development; these data can be downloaded from the fBIRN data repository website (http://fbirnbdr.nbirn.net:8080/BDR/).

The FBIRN Phase II study took place in 2005. Individuals with schizophrenia and healthy controls were imaged using structural and functional MRI imaging, and behavioral data, demographic information, and clinical assessments we also acquired on 253 subjects from around the US^6-8. Subjects were recruited locally, evaluated with a standardized clinical assessment battery, and scanned twice at each of the 9 participating FBIRN sites. A standardized protocol was used at each site which included the same cognitive (AO and SIRP) and calibration (SM and BH) fMRI tasks that were piloted during the fBIRN phase I traveling subjects study. As with the phase I study, while many scan parameters were matched across sites, important scan parameters were not the same across sites^6-8. Due to the large cohort of subjects imaged during phase II, analysis results from phase II data elucidated important differences between persons with schizophrenia and matched controls in the auditory oddball and working memory tasks as well in genetics and structural differences. The FBIRN collaboration has made the complete phase II data set to the scientific community for further methods development; these data can be downloaded from the BIRN data repository website (http://fbirnbdr.nbirn.net:8080/BDR/).

The third FBIRN study was the East Coast Traveling Subjects study (ECTS), which took place in 2007. Eighteen healthy participants were studied once at each of three magnet sites and twice at a fourth site. All sites were located on the East Coast of the United States. The purpose of this study was to provide a second dataset with which to assess test-retest and between-site reliability, to provide a rich data set to test tool and methods to allow for calibration of between site differences in FMRI results, and to evaluate how successful FBIRN site standardization methods were in reducing inter-site variability in fMRI data. The imaging protocol included a T1 weighted scan, seven runs of a working memory fMRI task, two calibration scans (breath hold (BH) and pulses arterial spin labeling (PASL)), and two resting state scans. Importantly, standardization methods that were identified and developed based on analysis from the previous two fBIRN studies were applied to this study. These included the use of a standard field strength (3T), us of vendor supplied receive only multi-channel head coils, and an EPI based k-space acquisition. In addition, a standard scanner check list was used during acquisition, scanner phantom QA data was acquired at all sites and metrics were actively monitored before and during data acquisition, human QA was performed during data acquisition, and a traveling engineer was sent to each site before the start of the study to provide final site evaluation and site standardization. Results from a multi-center analysis of the working memory data shows that between site variability is much smaller than between subject variability, suggesting that a dedicated effort of site standardization can generate a study with little residual inter-site variability ^{Brown 2011}. The FBIRN collaboration will make the complete ECTS data set available to the scientific community in 2011 on the BIRN data repository web site (http://fbirnbdr.nbirn.net:8080/BDR/).

The FBIRN West Coast traveling subjects study took place in 2008-9. Ten healthy participants and 10 participants with schizophrenia were studied twice at one site and once at each of two magnets at a second site, both located in Southern California. The purpose of this study was to provide a dataset s to evaluate test-retest and between-site reliability in both healthy participants and participants with schizophrenia. These data provide a rich data set to test tool and methods to allow for calibration of between site differences in FMRI results, and to evaluate how successful FBIRN site standardization methods were in reducing inter-site variability in fMRI data. The imaging protocol included the same imaging protocol as was used in the ECTS study. Methods developed by FBIRN to standardize multi-center fMRI studies were employed on each scanner. Analysis results from these multi-center data are ongoing and no results are available at this time. FBIRN collaboration will make the complete WCTS data set available to the scientific community in 2012 on the BIRN data repository web site (http://fbirnbdr.nbirn.net:8080/BDR/).

The final FBIRN study, phase III, collected data from 2010-11. Individuals with schizophrenia and healthy controls were imaged using a protocol including structural, perfusion, diffusion tensor, and fMRI scans. Behavioral data, clinical assessments, demographic data, and genetic data we acquired on all subjects. Subjects were recruited locally, evaluated with a standardized clinical assessment battery, and scanned once at each of the 7 participating FBIRN sites. The total recruitment goal was 210 participants with schizophrenia and 210 age and gender matched controls. The imaging protocol included a T1 weighted scan, seven runs of a working memory task, two runs of the phase II AO task, a PASL calibration scan, and resting state scans. As with the East and West coast traveling subjects studies, standardization of all aspects of the study was done to the extent possible. The large cohort of participants with schizophrenia and age and gender matched healthy are expected to elucidated additional important differences between subject populations and subpopulations within the cohorts in both the auditory oddball and working memory tasks, as well uncover important genetics and structural differences. Data analysis is currently underway. The FBIRN collaboration expects to make the complete phase III data set to the scientific community for further methods development as well as scientific investigations of schizophrenia; these data are expected to be made available in 2012 on the BRIN data repository website (http://fbirnbdr.nbirn.net:8080/BDR/).