Once a series of image registration methods have been used to bring all brain images in the study into alignment, the following variables are calculated and put in the 'processed' folder (see "Algorithm Overview"):
Jacobian - The Jacobian can be used to determine the factor by which brain regions in the set had to expand or contract in order to reach the average of all brains in the study. The Jacobian for a given voxel in the brain is a measure based on the deformation of each brain. It can be thought of as the amount by which the volume at that voxel had to be multiplied in order to match the average of all brains in the study. A Jacobian of 1 indicates the volume at that voxel matched the average of all brains; a Jacobian that is less than one indicates the volume at that voxel was greater than the average of all brains; a Jacobian that is greater than one indicates the volume at that voxel was less than the average of all brains. It is important to note that the Jacobian has been corrected for overall brain volume.
Scaled Jacobian - This is equivalent to the Jacobian, except it has not been corrected for overall brain volume. Therefore, the Scaled Jacobian is an absolute measurement while the Jacobian is a relative measurement.
Displacement - Displacement is measured by a vector extending from the initial position to the final position of a voxel. It measures the change in position between individual MRIs and the average brain.
Magnitude - Magnitude is the size of the displacement vector. A larger vector will have a larger magnitude. Likewise, smaller vectors will have smaller magnitudes.
Upon completion of the script, analysis of the data may be conducted using RMINC. The latest version of RMINC is available at the following website:
That website also contains a manual called VBMstats.pdf, which practically explains how to use RMINC. A second useful manual named RMINC.pdf is available at:
R must be able to locate where information about each brain is located (e.g. Jacobians, Scaled Jacobians, etc.). Other pertinent information may also be required for analysis such as Weight or Gender. First prepare a simple comma-separated file. In a spreadsheet, input the required data. Some examples of useful variables to include are Mouse ID Number, Weight, Genotype, Strain, Date of Birth, Gender, Image Number, etc. An example is given below:
It is important to make sure the labels are exactly the same (including white spaces and enter spaces), otherwise R will count them as separate categories. For example,
each 'GAMT' above was entered without any extra white spaces at the end. It is also important to be consistent with upper- and lowercase letters. For example, each 'Mutant' above begins with an uppercase letter. Once all information has been entered, save the file as a .csv file and make sure file name doesn't have white spaces.
Although the file above specifies the image name, it does not direct R to the required information about each image. The user needs to create a script which will print into a file full pathways for all required information. The script must specify level of blur. We recommend either a 1.0 or 0.5 blur when looking at total brain structure and volume. This level of blur provides enough detail for the purposes of looking at total brain. However, when analyzing individual brain structures, we recommend a 0.1 blur instead. A smaller amount of blur is required because, otherwise, blurring would run over the boundaries of the individual brain structures. The MICe centre is happy to forward an example of such a script upon request.
Once the 'Glim' file is ready, it can be read into R and the user is ready to begin analysis. RMINC is a library which is necessary to read and write MINC data into R. R package can then be used to run linear models, plot data, etc. Numerous types of analyses are possible in R. The manuals mentioned above are a great place to start as they work through practical examples, employing a variety of analyses.<!-- @page