MRI volume of the amygdala: a reliable method allowing separation from the hippocampal formation

Studies of MRI-derived volume of the amygdala have been mostly performed on coronal sections where its boundaries with the hippocampus and the entorhinal cortex are indistinct. To date, all reports of in vivo amygdala volume have consistently overestimated the size of the structure. We have developed a method for the MRI-based in vivo measurement of the amygdala volume which allows a better separation of the amygdala from the adjoining hippocampal formation. In nine normal volunteers we obtained three-dimensional spoiled gradient recalled acquisition, 1.3-mm thick, T1 weighted sagittal MR images and created electronically linked reformatted images in the coronal and axial planes. On the original sagittal and the reformatted axial planes, where it is more readily apparent, we delineated the boundaries between the amygdala and the hippocampus and the amygdala and the hippocampo-amygdala transition area, respectively. We then projected those markings onto the coronal plane, where the other boundaries of the amygdala are more easily seen. Using these markings as a guide and utilizing extra-amygdalar coronal landmarks for the anterior end, we outlined the whole amygdala on the coronal plane and determined its volume. We observed that 45% of the coronal slices that contained amygdala also contained some hippocampus. The amygdala measurement had high test-retest reliability, with an intra-class correlation coefficient (r(ICC)) of 0.99 for the total volume and an r(ICC) of 0.93 for the measurement at the level of the individual slice. The average amygdala volume was 1.05 +/- 0.17 cm(3) on the right and 1.14 +/- 0.15 cm(3) on the left. Our amygdala volumes are in agreement with those reported in postmortem studies, which provides the reported method with face validity. The ability to reliably and validly measure the amygdala in vivo may facilitate the investigation of psychiatric disorders, such as PTSD, anxiety disorders, and schizophrenia, which are presumed to have amygdala dysfunction and pathology. (C) 1999 Elsevier Science Ireland Ltd. All rights reserved.