S. L. Yates, A. Barach, S. Gingell, H. C. Whalley, D. Job, E. C. Johnstone, J. J. Best, S. M. Lawrie


0165-1781 (Print) 0165-1781 (Linking)

Publication year



Psychiatry Res

Periodical Number






Author Address

Division of Psychiatry, University of Edinburgh, Kennedy Tower, Royal Edinburgh Hospital, Morningside Park, Edinburgh EH10 5HF, UK.

Full version

A number of reliable techniques have been described that can parcellate temporal neo-cortex from MRI images to preserve topographical characteristics of individual brains, but these tend to use in-house software. We describe here an adaptation of the methods previously described by Kim et al. [Kim, J.J., Crespo-Facorro, B., Andreasen, N.C., O’Leary, D.S., Zhang, B., Harris, G., Magnotta, V.A., 2000. An MRI-based parcellation method for the temporal lobe. Neuroimage 11, 271-288], but utilising commercially and, therefore, generally available software. Using Analyze, we traced individual sulci and identified coronal bounding planes, and used a combination of three orthogonal plane views, manual limit tracing and semi-automated edge detection to parcellate 13 sub-regions of temporal neo-cortex from sets of serial coronal slices. We applied this technique to the baseline scans of the first seven subjects in the Edinburgh High Risk Study (EHRS) who developed schizophrenia, and a matched group of healthy controls, to see if temporal lobe sub-regional volumes could predict the onset of schizophrenia. Two relatively inexperienced raters developed these techniques in a short time period, and intra-rater intra-class correlation coefficients (ICC) ranged from 0.56 to 0.99, while the mean inter-rater ICC was 0.90 (range 0.55-0.99). There were, however, no significant differences in temporal lobe sub-regional volumes between the two groups we examined. We have, therefore, developed a reliable parcellation technique that requires relatively little training. It is, however, a laborious process, and it remains uncertain whether it is more sensitive to early disease processes in pre-schizophrenia than are other image-analysis techniques.