Computational protocol: Water diffusion closely reveals neural activity status in rat brain loci affected by anesthesia

Similar protocols

Protocol publication

[…] SPM8 software (Welcome Trust Centre for Neuroimaging, London, UK) and in-house software written in Matlab [] were used for data preprocessing and statistical analysis. Image preprocessing was performed first individually for each animal. Time-series images (DfMRI images with b1000 and b1800, and BOLD images) for each drug dosage were realigned to correct for residual head motion and coregistered to the reference structural images. Then, all images where spatially normalized and coregistered to the Paxinos and Watson rat brain atlases []. Finally, these images were resliced to a resolution of 0.2 × 0.2 × 0.2 mm3 and smoothed with a full width at half maximum (FWHM) Gaussian kernel of 0.6 mm.Time-series ADC maps were calculated separately for each animal at each drug dosage as ADC = ln(Sb1000/Sb1800)/800, in which Sb1000 and Sb1800 are the signal acquired with b = 1000 and 1,800 s/mm2, using the preprocessed time-series DfMRI images to remove any residual T2/T2* (BOLD) effects [], and averaged over the three directions to increase signal noise ratio and remove residual differences in gradient hardware (diffusion anisotropy effects were ignored in the regions of interest (ROIs) used for analysis at our spatial resolution). ADCs calculated with high b values (also called “synthetic ADC” or “shifted ADC” [sADC]), instead of typical values of 0 and 1,000 s/mm2, have more emphasis toward nonGaussian diffusion and water interaction with cell membranes []. Time-series BOLD signal maps for each animal at each drug dosage were normalized by the average value of the BOLD signal at a whole brain for each low dosage, respectively, as the BOLD signal is relative. Time-series ADC maps, which correspond to absolute measurements, were not normalized. The time-series maps of the absolute ADC values and normalized BOLD signals for each animal at each drug dosage were finally pooled for a group analysis. Then, a two-way repeated ANOVA analysis (cluster level corrected p < 0.05) was performed over the whole time courses and dosage conditions for isoflurane and medetomidine, respectively. Statistical ADC and BOLD maps were overlaid on the brain structural image. Overlay maps were created using MRIcron (http://www.cabiatl.com/mricro/mricro/mricro.html) to show regions commonly affected by isoflurane and medetomidine, for DfMRI and BOLD, respectively.ADC and BOLD time-course data were extracted for each drug dosage and for each animal from the clusters visible on the ADC overlay maps of (uncorrected p < 0.001), using MarsBar (MRC Cognition and Brain Sciences Unit, Cambridge, UK). Data were first averaged at each time point over all voxels of each ROI on an individual animal basis, then averaged for each ROI over all animals. The ROI number of voxels were 171 for SS, 58 for M, 79 for V, 180 for Au, 24 for Cg, 313 for CPu, 92 for Hip, 738 for Amy, 95 for Tha, 66 for HT, 58 for DR, and 112 for PAG, respectively. Abbreviations of the brain regions that were used in this report are shown in . ROIs corresponding to CM (21 voxels) and VPL (19 voxels) were manually drawn using MRIcron (). Time-series Sb1800 and Sb1000 signal maps after averaging over the three directions for each animal at each drug dosage were normalized by the average value of the Sb1800 and Sb1000 at a whole brain for each low drug dosage, respectively. Then, statistical Sb1800 and Sb1000 maps were also produced by two-way repeated ANOVA analysis (cluster level corrected p < 0.05) over the whole time courses and dosage conditions for isoflurane and medetomidine, respectively. Averaged Sb1800 and Sb1000 of each anesthetic dosage were calculated in the same way as the BOLD signal using the same ROIs.To provide some indication of the overall signal level in our image data, signal:noise ratios (SNRs) were estimated at whole brain for each animal and for each of the first five raw time-course images (before preprocessing and for each direction separately for the DfMRI data) under 0.1 mg/kg/h medetomidine. Averaged SNR values (mean ± SEM) were 8.51±0.25 for b1000 (n = 8), 5.57 ± 0.17 for b1800 (n = 8), and 11.32 ± 0.69 (n = 7) for BOLD. […]

Pipeline specifications

Software tools SPM, MRIcron
Application Magnetic resonance imaging
Organisms Rattus norvegicus
Chemicals Furosemide