Automated Talairach atlas-based parcellation and measurement of cerebral lobes in children

• Published in: Psychiatry research Vol. 91; no. 1; pp. 11 - 30
• Main Authors: Kates, Wendy R., Warsofsky, Ilana S., Patwardhan, Anil, Abrams, Michael T., Liu, Alex M.C., Naidu, SakkuBai, Kaufmann, Walter E., Reiss, Allan L.
• Format: Journal Article
• Language: English
• Published: Ireland Elsevier Ireland Ltd 1999
• Subjects: Adult; Brain - anatomy & histology; Brain - pathology; Brain parcellation; Brain Stem - pathology; Case-Control Studies; Cerebellum - pathology; Child; Female; Frontal Lobe - pathology; Humans; Image Processing, Computer-Assisted; Magnetic Resonance Imaging; Male; Models, Anatomic; Occipital Lobe - pathology; Parietal Lobe - pathology; Pediatric; Positive predictive value; Reference Standards; Reference Values; Rett Syndrome - pathology; Sensitivity; Sensitivity and Specificity; Specificity; Temporal Lobe - pathology; Brain parcellation; Sensitivity; Specificity; Positive predictive value; Pediatric
• ISSN: 0925-4927; 0165-1781; 1872-7506; 1872-7123
• DOI: 10.1016/S0925-4927(99)00011-6

This study applied a Talairach-based automated parcellation method, originally proposed for adults, to the measurement of lobar brain regions in pediatric study groups. Manual measures of lobar brain regions in a sample of 15 healthy boys, girls and adults were used initially to revise the original Talairach-based grid to increase its applicability to pediatric brains. The applicability of the revised Talairach grid was then tested on an independent sample of five girls with Rett syndrome. As Tables 3 and 4 in the text demonstrate, sensitivity, specificity and positive predictive values either remained unchanged or increased as a result of revising the sectors to fit the brains of children. High levels of sensitivity and specificity were achieved for all revised Talairach-based calculations in relation to the manual measures. Both positive predictive values and intraclass correlations between volumetric measures produced by the revised automated and manual methods varied with the relative size of the brain region. Values were relatively low for smaller structures such as the brainstem and subcortical region, and high for lobar regions. These results suggest that the automated Talairach atlas-based parcellation method can produce sensitive and specific volumetric measures of lobar brain regions in both normal children and children with brain disorders. Accordingly, the method holds much promise for facilitating quantitative pediatric neuroimaging research.