deletions | additions       diff --git a/Assumption_is_made_that_the__.md b/Assumption_is_made_that_the__.md  index b39c9a5..878cc2e 100644  --- a/Assumption_is_made_that_the__.md  +++ b/Assumption_is_made_that_the__.md 
...
* Assumption is made that the major contribution to the boundary variability are the contrast differences and gradient distortions, but it was shows (see Tardif 2010 "Regional Impact of Field Strength on Voxel-Based Morphometry Results" and Tardif 2009 "Sensitivity of voxel-based morphometry analysis to choice of imaging protocol at 3 T"), that different protocols and field strengths significantly affect signal and contrast inhomogeneity patterns: which leads to differences in regional accuracy of tissue classification due to different contrast to noise ratio.  The paper cited here add adds  evidence for our hypothesis that different ROIs will have different scaling factors, due to regional differences in contrast, as shown in a VBM analysis by Tardiff and colleagues. I have edited this section:   *We hypothesized that all differences in* ***regional*** *contrast and geometric distortion result in regional volumes that are consistently and linearly scaled from their true value. For a given region of interest (ROI), two mechanisms impact the final boundary definition: gradient nonlinearities cause distortion and, simultaneously, hardware (including scanner, field strength, and coils), and acquisition parameters modulate tissue contrast, adjusting the whole boundary.* ***Previously, Tardiff and colleagues have found that contrast-to-noise ratio and contrast inhomogeneity from various pulse sequences and scanner strengths cause regional biases in VBM\cite{tardif2010regional,tardif2009sensitivity}, and therefore we hypothesized that each ROI will scale differently at each site.*** *By imaging 12 subjects in 20 different scanners using varying acquisition schemes, we were able to estimate the scaling factor for each regional volume at each site...*