In vivo basal ganglia volumetry through application of NURBS models to MR images

Volumetry of basal ganglia (BG) based on magnetic resonance imaging (MRI) provides a sensitive marker in differential diagnosis of BG disorders. The non-uniform rational B-spline (NURBS) surfaces are mathematical representations of three-dimensional structures which have recently been applied in volumetric studies. In this study, a volumetric evaluation of BG based on NURBS was performed in 35 right-handed volunteers. We aimed to compare and validate this technique with respect to manual MRI volumetry and evaluate possible side differences between these structures. Intra- and interobserver biases less than 1.5% demonstrated the method's stability. The mean percentage differences between NURBS and manual methods were less than 1% for all the structures considered; however, the internal segments of the globus pallidus showed a mean percentage difference of about 1.7%. Rightward asymmetry was found for the caudate nucleus ( mean +/- SD 3.20 +/- 0.20 cm(3) vs. 3.10 +/- 0.19 cm(3), P< 0.001) for both its head (1.44 +/- 0.10 cm(3) vs. 1.41 +/- 0.09 cm(3), P< 0.01) and its body/tail (1.73 +/- 0.11 cm(3) and 1.68 +/- 0.12 cm(3), P< 0.01), and for the globus pallidus (1.23 +/- 0.08 cm(3) and 1.18 +/- 0.09 cm(3), P< 0.001) for both the internal (0.33 +/- 0.05 cm(3) vs. 0.31 +/- 0.05 cm(3), P< 0.01) and external (0.90 +/- 0.05 cm(3) vs. 0.86 +/- 0.05 cm(3), P< 0.001) segments. No volumetric side differences were found for the putamen (3.43 +/- 0.14 cm(3) vs. 3.39 +/- 0.17 cm(3), P> 0.05). The rightward asymmetry of the BG may be ascribed to the predominant use of the right hand. In conclusion, NURBS is an accurate and reliable method for quantitative volumetry of nervous structures. It offers the advantage of giving a three-dimensional representation of the structures examined.