Exemplar 1 (dementia)
Prospective study to assess Grid-based services to improve consistency and analysis of MR imaging in longitudinal studies and trials
Aim: To assess Grid-based services to improve consistency of MR imaging of atrophy in longitudinal studies and trials and assess the Grid service concept for image analysis.
Worlplan: 30 subjects with Alzheimer’s disease (AD), 15 subjects with Mild Cognitive Impairment (MCI) and 30 age-matched normal elderly controls will be recruited from ongoing longitudinal studies currently underway at the Institute of Neurology (IoN, UCL), Cambridge and Newcastle. All subjects will have full clinical, diagnostic and neuropsychometric workup and will be consented to the extra MR studies at 0, 6 and 12 months. Scans will be transferred from the MR system following acquisition and uploaded onto the Grid. Linked demographic, clinical and neuropsychological data will also be uploaded at each site and stored along with the images in the exemplar database. In addition, subjects from the ION site (10 AD, 5 MCI and 10 controls) will have second scans, performed on the same day, at the Hammersmith Hospital. Same-day scans using the same protocol with the same patients will allow differences due to different scanners to be assessed, and enable validation of normalisation methods devised in level 2 to adjust for differences between scanners. Half of the subjects recruited at each site (randomised balanced allocation) will have the NeuroGrid quality assurance intervention (level 2 toolkit) for their follow-up scans: each individual’s MR imaging at 6 and 12 months will be assessed for adequacy of acquisition and for consistency with their earlier scans. This will allow for real-time correction of acquisition problems and, if necessary, re-scanning before the patient has left the MR department. The effects of this intervention will then be assessed in comparison with subjects scanned with normal QA procedures. The level of movement artefact, grey-white contrast and signal to noise will be assessed on all scans blind to whether there has been Grid-based intervention. Changes in contrast or inhomogeneity will also be quantified using a Grid registration service of follow-up scans to baseline. Rates of atrophy and ventricular enlargement will be measured on all subjects using both the NeuroGrid image analysis toolkit, and also traditional semi-automated image analysis tools. The variance of each measure for each subject group will be assessed, together with correlations with clinical and neuropsychological tests. The hypothesis is that improved consistency of acquisition should reduce variance in the MR-derived measures within each subject group (both within- and across-scanners) and thereby improve correlation with clinical measures. Improvements in variance and clinical correlations would be of great value to future trials with MRI outcome measures.
Contribution to the development of NeuroGrid: This exemplar requires real-time transfer and processing of images to return information to the scanner before the examination is over so that data quality can be assured and, if necessary, there can be an intervention to deal with problems. It will use both secure data transfer and Grid services. From the augmented studies, there will be scientific outcomes to demonstrate added value resulting from the Grid; specifically, from the direct benefits of the Grid toolkit to reduce variance resulting from multiple sites and also from cooperative working and aggregation of data.
