Matlab 8

Image analysis was carried out with SPM5 software (Wellcome Department of Imaging Neuroscience, London, UK; www.fil.ion.ucl.ac.uk/spm) on MATLAB 8 (MathWorks Inc, Sherborn, Mass). At the single subject level, ¹⁸F-FDG-PET imaging analysis has been performed according to an optimized SPM FDG-PET pipeline previously developed and validated [62 ]. In a first step, scans were ‘spatially normalized’ in accordance to a reference FDG-PET “dementia- specific” template [63 (link)].
In a second step, spatially normalized and smoothed images for a single patient were then compared to a large group of control scans by means of a two-sample t-test implemented in SPM5 to assess areas of hypometabolism throughout the whole-brain at a single-subject level. Proportional scaling was used to remove intersubject global variation in PET intensities. The threshold for assessing hypometabolism was set at p = 0.05, FWE-corrected for multiple comparisons at the voxel level. Only clusters containing more than 100 voxels were deemed to be significant. The resulting single-subject SPM hypometabolic maps have been also visually inspected by a team of experts (neurologists, radiologists, and nuclear medicine physicians) in PET imaging in order to further validate svPPA pathologically hypometabolic areas in each patient.

The fusion of the data obtained from both ionization modes, i.e., ESI⁺ and ESI⁻, was carried out by concatenation after autoscaling and a normalization step based on the Frobenius norm of each dataset. AMOPLS was computed under the MATLAB^® 8 environment (The MathWorks, Natick, MA, USA). To avoid detrimental effects on interpretation due to unbalanced groups, a stratified subsampling strategy without replacement was carried out. Differences between glass aquaria were considered negligible, while the combination of gender and exposure experimental factors was used for the random selection of 9 mussels in each experimental group (i.e., the smallest size of exchangeable units). By these means, a population of 10³ balanced AMOPLS models was generated and parameters, including sum-of-squares, scores, and loadings, were then investigated to offer robust ensemble-based estimates. Additional details about this procedure can be found in Boccard et al. [52 (link)].

The MRI data were analyzed using SPM8 software (Welcome Trust Centre for Neuroimaging, London, UK; http://www.fil.ion.ucl.ac.uk/spm) implemented in Matlab 8 (The MathWorks, Natick, MA). All volumes were coregistered to the SPM single-subject canonical EPI image, slice-time corrected and realigned to the mean image volume. A high resolution anatomical image of each subject was first coregistered to the SPM single-subject canonical T1 image and then to the average functional image. The transformation matrix obtained by normalizing the anatomical image was used to normalize functional images to MNI space. The normalized images were spatially smoothed using an 8-mm full width at half maximum Gaussian kernel. A high-pass temporal filter with cutoff of 432 s and 512 s, respective to the short or long session, was applied to remove low-frequency drifts from the data.