Using a structural ROI

So far we have used a functional ROI. This has the advantage that it is usually well tuned to the subject we are analysing. The disadvantages are that we have had to use a whole run of data to define the ROI, which we would have preferred to be able to analyze, and that functional ROIs can be noisy, when the activation signal is not strong. An alternative is to use the anatomy of the brain to estimate the location of functional areas.

Using anatomical ROIs can work well for areas that are naturally defined by brain structure, such as the subcortical nuclei, or the primary sensory and motor cortices, where the functional areas are closely linked to the position of large and relatively invariant sulci. Outside these areas, it can be difficult to define functional areas using anatomy alone. The problems are compounded when anatomical ROIs are defined on one subject, and applied to another, because there is great variability between subject in sulcal anatomy.

In the example experiment, subjects responded with a key-press each time they saw the flashing checker board. We might therefore be interested to know the level of activation in the putamen. This would be a good candidate for an anatomical ROI, because the putamen can be accurately defined on a structural scan, and does not vary much between subjects after spatial normalization. The AAL ROI library contains a definition of the left and right putamen for a single subject after spatial normalization. The images from our subject have been spatially normalized, so the AAL structural ROIs definition of the putamen will probably give a reasonable approximation to the putamen in our data.

Running an analysis using structural ROIs

is exactly the same as running the analysis with the functional ROI. Select Design from the MarsBaR menu, and Set design from file. Choose sess1/SPM8_ana/SPM.mat. Click on Data, Extract ROI data (default). When you are asked for ROI file, navigate to the MarsBaR Example dataset directory, then to the rois subdirectory, select MNI_Putamen_L_roi.mat and click Done. This is a copy of one of the AAL structural ROIs. When the data extraction is done, choose Results, Estimate results and wait till MarsBaR has done its thing. Select Results, Statistic table, enter the stim_hrf contrast, as shown in the figure above. Repeat the same procedure, using the copy of the AAL MNI_Putamen_R_roi.mat ROI. You will now have two tables.

One table for the left putamen:

Contrast name    ROI name: Contrast value:    t statistic:  Uncorrected P:    Corrected P
-----------------------------------------------------------------------------------------

stim_hrf
---------------------------

                Putamen_L:           0.08:           0.77:       0.222894:       0.222894

and one for the right putamen:

Contrast name    ROI name: Contrast value:    t statistic:  Uncorrected P:    Corrected P
-----------------------------------------------------------------------------------------

stim_hrf
---------------------------

                Putamen_R:           0.05:           0.50:       0.307312:       0.307312

The subject responded with their right hand, so we expected that the right putamen would have less signal than the left.

Batch mode

You can also run MarsBaR in batch mode. There is an example batch script in the <marsbar>example/batch directory, called run_tutorial.m. You won’t be suprised to hear that this is a batch script that runs most of the steps in this tutorial, as well as extracting and plotting reconstructed event time courses.

The end

That is the end of this short guided tour. We haven’t described the options interface, but then again, it isn’t very interesting. As always, we would be very grateful to hear about any mistakes in this document or bugs in MarsBaR. You can find us on the MarsBaR Discourse site - see Support.

May your regions always be as interesting as you hoped.

The MarsBaRistas