A dual echo coronal sequence is processed using REGION to obtain the following volumes: intracranial, total brain, ventricular cerebrospinal fluid (CSF), subarachnoid CSF, frontal lobe, parietal-occipital region, basilar region, white matter high signal, and left and right hemispheric breakdowns for all of the above. Our most involved procedure, full coronal analysis, requires six to eight hours to perform. Roughly, the steps are as follows:
1) The dual coronal sequence is imported into REGION.
2) A sample file is created by selecting representative pixels of each of four tissue types throughout the brain: brain tissue, fluid, bone, white matter high signal. REGION will use the multi-echo intensity and location data from each sample pixel to perform linear regressions to best classify tissue types.
3) An automated algorithm strips away the skull. The user follows atlas-based landmarks to correct any errors in the skull removal and to remove the cerebellum.
4) A semi-automated algorithm is applied to each slice to differentiate tissue types and quantify the amounts of pixels for each.
5) Further divisions are made by the user to distinguish lobar sub-volumes, using atlas-based rules.
6) The pixel areas for each region of interest (R.O.I) are summed and converted to cubic centimeters.
7) The central sagittal image is imported to NIH Image. The length of the frontal lobe is measured to correct for the lobar boundary (the central sulcus) which is not viewable on the coronal images.

The image below illustrates the many regions of interest measured on just one slice of a coronal set. The same algorithm is repeated for every image in the brain, often as many as 45-50 slices.