We are sorry, but NCBI web applications do not support your browser and may not function properly. More information

Results: 4

1.
Fig. 4

Fig. 4. From: False positives in neuroimaging genetics using voxel-based morphometry data.

12 mm VBM image non-stationarity. The figure illustrates the variation in image smoothness measured in FWHM, derived from the SPM RPV image. There is a wide variation, ranging from 3.8 to 27.7 mm.

Matt Silver, et al. Neuroimage. 2011 January 15;54(2):992-1000.
2.
Fig. 3

Fig. 3. From: False positives in neuroimaging genetics using voxel-based morphometry data.

Distribution of voxel-wise FWHM for ADNI images smoothed with 6 mm (left) and 12 mm (right) Gaussian smoothing kernels. Voxel-wise FWHM gives an indication of local smoothness and corresponds to the ‘full-width at half-maximum’ of a Gaussian kernel required to produce a random (white noise) image of equivalent smoothness. A perfectly stationary image would have constant FWHM at all voxels. In contrast, a highly non-stationary image would have a large spread in FWHM, as is seen here.

Matt Silver, et al. Neuroimage. 2011 January 15;54(2):992-1000.
3.
Fig. 1

Fig. 1. From: False positives in neuroimaging genetics using voxel-based morphometry data.

Non-stationary image simulation. (a) Schematic illustrating extent of 3 different smoothness regions. (b) as (a) with ADNI image brain mask applied. (c) Realisation of non-stationary image with outer, middle and inner regions smoothed with 8, 12 and 18 mm FWHM Gaussian smoothing kernels. (d) as (c) with final 1.5 mm smoothing kernel and ADNI mask applied.

Matt Silver, et al. Neuroimage. 2011 January 15;54(2):992-1000.
4.
Fig. 2

Fig. 2. From: False positives in neuroimaging genetics using voxel-based morphometry data.

Accuracy of estimation of c, the theoretical number of clusters under RFT. Histograms show the empirical distribution of c across all 700 SNPs at three different cluster-forming thresholds (left to right), and with two different smoothing kernels (top and bottom). The theoretical (RFT) and empirical mean number of clusters, , are shown as dashed and dotted lines respectively. The amount by which RFT overestimates increases as the cluster-forming threshold uc is lowered, and with images of lower smoothness. (Note that the x axis for 6 mm smoothed images has a larger range, reflecting the fact that many more clusters are observed).

Matt Silver, et al. Neuroimage. 2011 January 15;54(2):992-1000.

Supplemental Content

Recent activity

Your browsing activity is empty.

Activity recording is turned off.

Turn recording back on

See more...
Write to the Help Desk