Recent developments in three-dimensional (3D) Magnetic Resonance Imaging (MRI) open new research horizons. These non-invasive techniques afford the possibility of studying in vivo cerebral structures of both patients and healthy subjects. Furthermore, MRIs gives an operational definition of cerebral structures. A study of the relation between variation in one or several cerebral structures and variation in behavior could be considered in neuropsychology. For this purpose, biometric qualities, i.e. reliability and validity, are required for cerebral volumetric measurements. Using MRI, 124 contiguous slices of the brain can be obtained and the brain can then be reconstructed in 3D for surface area or volume evaluations.

In this study, images of brains from 10 healthy females aged 20.15 years (± 1.14 years) were analyzed to perform cerebral volume evaluations. The results of four techniques were compared. One is based on manual contour of the brain, two are semi-automatic and one is automatic. The cerebral volumes obtained by the two semi-automatic techniques did not differ statistically (1233 ± 86.2 cm³ and 1218 ± 110.5 cm³), nor did they differ when the manual and automatic techniques were compared (1305 ± 100.6 cm³ and 1311 ± 82.3 cm³). On the other hand, the two latter techniques gave volumes statistically different from those given by the two other. The intra-class correlation coefficients (ICC) in assessing inter-rater reliability were over 0.97. For correlation between techniques, three ICC were below 0.66 and not statistically significant. Only the manual contour and automatic techniques showed a significant correlation, giving a proportion of common variance (92%) that was satisfactory but not biometrically perfect. Although reliable, these techniques are not interchangeable.

Neuropsychologists are interested in the relationship between behavioral variation and cerebral structures. This relationship could be studied from raw values of cerebral structure or from adjusted measures on brain volume. In order to compare results of studies using adjusted measures, a single and relevant evaluation technique of brain volume which will be the same for all researches is necessary.