The Italian Alzheimer's Disease Neuroimaging Initiative (I-ADNI): validation of structural MR imaging.

BACKGROUND: The North American Alzheimer's Disease Neuroimaging Initiative (NA-ADNI) was the first program to develop standardized procedures for Alzheimer's disease (AD) imaging biomarker collection. OBJECTIVE: We describe the validation of acquisition and processing of structural magnetic resonance imaging (MRI) in different Italian academic AD clinics following NA-ADNI procedures. METHODS: 373 patients with subjective memory impairment (n = 12), mild cognitive impairment (n = 92), Alzheimer's dementia (n = 253), and frontotemporal dementia (n = 16) were enrolled in 9 Italian centers. 22 cognitively healthy elderly controls were also included. MRI site qualification and MP-RAGE quality assessment was applied following the NA-ADNI procedures. Indices of validity were: (i) NA-ADNI phantom's signal-to-noise and contrast-to-noise ratio, (ii) proportion of images passing quality control, (iii) comparability of automated intracranial volume (ICV) estimates across scanners, and (iv) known-group validity of manual hippocampal volumetry. RESULTS: Results on Phantom and Volunteers scans showed that I-ADNI acquisition parameters were comparable with those one of the ranked-A ADNI scans. Eighty-seven percent of I-ADNI MPRAGE images were ranked of high quality in comparison of 69% of NA-ADNI. ICV showed homogeneous variances across scanners except for Siemens scanners at 3.0 Tesla (p = 0.039). A significant difference in hippocampal volume was found between AD and controls on 1.5 Tesla scans (p < 0.001), confirming known group validity test. CONCLUSION: This study has provided standardization of MRI acquisition and imaging marker collection across different Italian clinical units and equipment. This is a mandatory step to the implementation of imaging biomarkers in clinical routine for early and differential diagnosis.

Spectroscopic magnetic resonance imaging of the brain: voxel localisation and tissue segmentation in the follow up of brain tumour.

The field of application of magnetic resonance spectroscopy (MRS) in biomedical research is expanding all the time and providing opportunities to investigate tissue metabolism and function. The data derived can be integrated with the information on tissue structure gained from conventional and non-conventional magnetic resonance imaging (MRI) techniques. Clinical MRS is also strongly expected to play an important role as a diagnostic tool. Essential for the future success of MRS as a clinical and research tool in biomedical sciences, both in vivo and in vitro, is the development of an accurate, biochemically relevant and physically consistent and reliable data analysis standard. Stable and well established analysis algorithms, in both the time and the frequency domain, are already available, as is free commercial software for implementing them. In this study, we propose an automatic algorithm that takes into account anatomical localisation, relative concentrations of white matter, grey matter, cerebrospinal fluid and signal abnormalities and inter-scan patient movement. The endpoint is the collection of a series of covariates that could be implemented in a multivariate analysis of covariance (MANCOVA) of the MRS data, as a tool for dealing with differences that may be ascribed to the anatomical variability of the subjects, to inaccuracies in the localisation of the voxel or slab, or to movement, rather than to the pathology under investigation. The aim was to develop an analysis procedure that can be consistently and reliably applied in the follow up of brain tumour. In this study, we demonstrate that the inclusion of such variables in the data analysis of quantitative MRS is fundamentally important (especially in view of the reduced accuracy typical of MRS measures compared to other MRI techniques), reducing the occurrence of false positives.