A Comparison of Reported Hemodynamic Activation in SPM2b vs. SPM99 - Kim Celone

Image processing and data analysis were preformed on a dataset collected from UBC in both spm99 and spm2b. The images that follow display that there is close to a 10% increase in reported hemodynamic activity when processing image data using spm2b rather than spm99.

Fig 1. SPM99 Random effects analysis showing target activations at p < .001 uncorrected

Fig 2. SPM2b Random effects analysis showing target activations at p < .001 uncorrected (intrinsic autocorrelation was also enabled)

NOTES ON THE ANALYSIS

Reorientation procedures appear to remain the same. There are several new variables within the realignment and normalization that are discussed below. A listing of the new SPM2b features and functionality can be found at:
http://www.fil.ion.ucl.ac.uk/spm/spm2b.html#New

Realignment Parameters:

Registration Quality: Quality versus speed trade-off. Highest quality (1) gives most precise results, whereas lower qualities gives faster realignment. The idea is that some voxels contribute little to the estimation of the realignment parameters. This parameter is involved in selecting the number of voxels that are used. You will most likely use 1.

Allow Weighting of Reference Image: Give the option of providing a weighting image to weight each voxel of the reference image differently when estimating the realignment parameters. The weights are proportional to the inverses of the standard deviations. For example, when there is a lot of extra-brain motion – e.g., during speech, or when there are serious artifacts in a particular region of the images. You will most likely use ‘no weighting’.

Reslice Interpolation Method: The method by which the images are sampled when being written in a different space.
‘Nearest Neighbour’ – Fastest, but not normally recommended.
‘Bilinear Interpolation’ – OK for PET, or realigned fMRI.
‘B-Spline Interpolation’ – Better quality (but slower) interpolation, especially with higher degree splines. Don’t use B-splines when there is any region of NaN of Inf in the images. You will most likely use ‘B-spline Interpolation’.
‘Fourier Space Interpolation’ – Rigid body rotations are executed as a series of shears, which are performed in Fourier space (Eddy et. Al. 1996). Unfortunately, this method can only be applied images with cubic voxels (since zooms can not be done by convolution in Fourier space).

Wrapping: No wrapping will most likely be used at the NRC.
‘No Wrapping’ – for PET or images that have already been spatially transformed.
‘Wrap in Y’ – for (un-resliced) MRI where phase encoding is in the Y direction (voxel space).

Normalization Parameters:

Weighting template when registering – Applies to weighting mask to the template(s) during the parameter estimation. With the default brain mask, weights in and around the brain have values of one whereas those clearly outside the brain are zero. This is an attempt to base the normalization purely upon the shape of the brain, rather than the shape of the head (since low frequency basis functions can not really cope with variations in skull thickness).
The option is now available for a user specified weighting image. This should have the same dimensions and mat file as the template images, with values in the range of zero to one.
‘no weighting’ – no template weighting
‘Default Brain Mask’ – weighting with …/apriori/brainmask.mnc
‘Specified Weighting’ – user selected weighting file.

Cutoff – Cutoff of DCT bases. Only DCT bases of periods longer than the cutoff are used to describe the warps. The number used will depend on the cutoff and the field of view of the template image(s). 25mm cutoff provides 4 x 5 x 4 basis function.

Bounding Box – The bounding box (in mm) of the written normalized images. Default is most often sufficient.

*Interpolation method and Wrapping are defined the same as when used in realignment.

Compatibility…between SPM2b and SPM99 (an important note from the SPM2b website)

One thing to watch out for is the image orientation. The proper Analyze format uses a left-handed co-ordinate system, whereas Talairach uses a right-handed one. In SPM99, images were flipped at the spatial normalisation stage (from one co-ordinate system to the other). In SPM2b, a different approach is used, so that either a left- or right-handed co-ordinate system is used throughout. The SPM2b program is told about the handedness that the images are stored with by the spm_flip_analyze_images.m function and the defaults.analyze.flip parameter that is specified in the spm_defaults.m file. These files are intended to be customised for each site. If you previously used SPM99 and your images were flipped during spatial normalisation, then set defaults.analyze.flip=1. If no flipping took place, then set defaults.analyze.flip=0.
Check that when using the Display facility (possibly after specifying some rigid-body rotations) that:
• The top-left image is coronal with the top (superior) of the head displayed at the top and the left shown on the left. This is as if the subject is viewed from behind.
• The bottom-left image is axial with the front (anterior) of the head at the top and the left shown on the left. This is as if the subject is viewed from above.
• The top-right image is sagittal with the front (anterior) of the head at the left and the top of the head shown at the top. This is as if the subject is viewed from the left.

***All Parameter descriptions were retrieved from the SPM2b Help Documentation***