[Optimization and diagnostic accuracy of computerized tomography with tridimensional spiral technique in the study of craniostenosis].

INTRODUCTION: Conventional Computed Tomography (CT) with three-dimensional (3D) reconstructions is considered the most complete and accurate imaging modality to diagnose craniosynostosis. However, the introduction of Spiral CT (SCT) opened new possibilities for 3D studies of the skull in pediatric patients with craniosynostosis. The purpose of our study is two fold: first, to optimize the scanning and imaging parameters to obtain diagnostic images in a single spiral scan; second, to assess the diagnostic accuracy of such images in the identification of normal and abnormal cranial vault sutures.

MATERIAL AND METHODS: Seventy-eight pediatric patients (age range: 1-35 months; mean: 11.8 months) with craniosynostosis were submitted to SCT of the head. The images were acquired with the following parameters: 3- and 5-mm nominal slice thickness, 5-6 mm/s table feed (pitch 1-2), 165 mAs and 120 kV. Two different algorithms and increases were used for image reconstructions. A first set of images was reconstructed with 2-mm increases and a soft tissue algorithm: these images were used for brain studies and for 3D reconstructions. A second set of slices was reconstructed with 5-mm increases and a bone algorithm to visualize the sutures of the axial plane. The 3D images were processed with the Shaded Surface Display software with threshold values ranging 120-150 HU. All images were acquired with a single spiral scan lasting less than 30 seconds. Two blinded radiologists analyzed the 3D and the planar images independently to evaluate the course and depth of each cranial suture. The sensitivity, specificity and diagnostic accuracy of both 3D and planar SCT images were evaluated. The frequency of artifacts (the Lego effect, boiled egg, pseudoforamina, movement, and chainsaw artifacts) and their influence on the final diagnosis were studied on 3D SCT images.

RESULTS: The diagnostic accuracy rates of 3D SCT images, by suture, were: sagittal 90.7%, metopic 100%, left lamboid 90.9%, right lamboid 93.9%, left coronal 85.7%, right coronal 91.1%. The diagnostic accuracy rates of the axial images, by suture, were: sagittal 90.7%, metopic 95.5%, left lamboid 86.4%, right lamboid 90.9%, left coronal 83.7%, right coronal 91.1%. The interobserver agreement on 3D images was: sagittal 91.1%, metopic 100%, left lamboid 88.9%, right lamboid 91.1%, left coronal 88.9%, right coronal 84.4%. The Lego effect artifact was the most frequent one (82%) and affected image evaluation in 6.3% of cases.

CONCLUSIONS: Our results prove that 3D SCT is a very accurate technique for identifying normal and abnormal sutures and presents many advantages over conventional 3D CT in the examination of pediatric patients with craniosynostosis. The quality of 3D SCT images was adequate and the artifacts did not affect the final diagnostic yield significantly.