The first step to process your infant MRI data using iBEAT is to perform image preprocessing. This step includes 1) reorientation and resampling and 2) N3 correction. To start the image preprocessing, click the Image Preprocessing button in the box that popped up when you typed “ibeat” into the terminal. Then, load in your subject images following the instructions provided in iBEAT: Loading Subject Images.

The goal of the reorientation and resampling process is to put your raw data into the correct orientation so that the sagittal, axial, and horizontal views all resemble the template. Once you load in your subject images, you can load in a template by clicking the “Load/Show Template” button at the top of the image preprocessing window. This will open a new window that shows you the sagittal, axial, and horizontal views of a template. You can select a neonate, 1-year-old, or 2-year-old template from iBEAT, or load in your own custom template. If you compare the template views to your raw data, look at whether the sagittal view is correctly oriented (i.e., nose facing to the right), whether the axial view is correctly oriented (i.e., top of head at the top of the box, neck and shoulders at the bottom of the box), and whether the horizontal view is correctly oriented (i.e., front of brain at the top of the screen, back of brain at the bottom of the screen). If the images match the template orientation, you can set your orientation parameters to x, y, z and click “reorient and resample.” If they do not, you have two options:

1. If you can easily identify the change that needs to be made (e.g., nose is facing left in sagittal view, but all other views are correct) you can change the orientation parameters to make the change, and click “reorient and resample.”

1. If you cannot easily identify the change that needs to be made (e.g., nose is facing up in the sagittal view, top of the head is at the right side of the box in the axial view, etc.), you can right click on the image name in the list of subject images (e.g., “S001-0-T1”) and click “Review Reorientation Options.” This will open a new window that shows all of the possible reorientation parameters and how those reorientation parameters will change the images. You want to find the image in this list that best matches the template, which is shown at the top of the box. There is a scroll bar, so be sure to scroll down and choose the image that exactly matches the orientation of the template.

You can reorient multiple images at once if the images all have the same reorientation parameters. Just select “T1,” “T2,” or “FA,” from the drop-down menu if you want to reorient all images of one type at once. If you want to reorient all images, select “All.” For this to work correctly, all of the images must have the same reorientation parameters. For this reason, I tend to reorient each image individually (by choosing “Selected” from the drop-down menu). Reorienting individually ensures that the image has the correct reorientation parameters.

Once you reorient and resample, a new window will open with the corrected image. You can compare this corrected image to ensure that it matches the template image. If not, you can always try again. A new file will also be listed under subject images (e.g., S001-0-T1-reoriented) and will appear in your subject’s folder for that timepoint.

One problem I have noticed in applying the reorientation and resampling to my own data is that when data is collected from infants, their heads are not always aligned well in the scanner. In our study, we collect data from naturally sleeping infants (i.e., their parents help them to fall asleep in the lab using a normal bedtime routine, and we place them into the scanner once they have fallen asleep). Because we are worried about waking the sleeping infants, we tend not to readjust their heads in the scanner if they are not perfectly straight. This can lead to some images with sagittal and axial views that look quite similar. The researcher must use his or her best judgment as to which image is the sagittal view and which image is the axial view. As long as the researcher can make a reasonable judgment, problems with head alignment in the scanner do not seem to be a major problem during this step. They may lead to problems during tissue segmentation, which will be discussed in that section. I believe a good solution would be to align the brain with the templates through another software (e.g., FSL, SPM) before bringing the images into iBEAT. I will describe this alignment step in another post.

Once you have completed the reorientation and resampling for all images, you can perform N3 correction. If you try to apply N3 correction before you have reoriented and resampled all of your images, a box will pop up warning that you should first complete the reorientation and resampling of all of your images. If you have already reoriented and resampled all of your images, click “N3 Correction on All Processed Images.” This will adjust the intensity of your images so that there are not brighter and darker spots throughout the image. The output will be the same reoriented images (e.g., “S001-0-T1-reoriented”) created in the previous step. Therefore, if anything goes wrong, you should go back to the reorient and resample step and start over.

Often, the output of the N3 correction will be images that are slightly darker than the original images. This is correct, but may make the images a bit harder to see. iBEAT will adjust the brightness of the image again in the brain extraction step so that it is easier to see. For now, as long as the green cross-hairs appear when you click on the reoriented image, you have completed the N3 correction. When there are more exteme intensity problems, the image may look so dark that you can no longer see the brain. Again, this will be fixed in the brain extraction stage, so don’t worry about it for now. Once you are satisfied that N3 correction worked for all of your images, you can click “Brain Extraction,” which will close the image preprocessing window and open the brain extraction window with your subject images already loaded.