Reproducibility assessment of Delaire cephalometric analysis using reconstructions from computed tomography

Cephalometry performed on sagittal teleradiographic X-rays is commonly used for the planification of orthognathic procedures. Tomographic imaging techniques (computer tomography or cone-beam computer tomography) are more and more prescribed for the assessment of complex occlusofacial anomalies. In this study, we intended to evaluate if teleradiographic X-rays reconstructed from computer tomography (CT) can be used to trace reliable cephalometric analyses. We assessed the inter- and intra-observer reproducibility of a simplified Delaire analysis (Top 12), one of the numerous common cephalometric analyses, performed with the Delaire Evolution software on three sets of X-rays: (1) conventional cephalometric X-rays, (2) 2D reconstructions obtained from the CT scans of the same patients and (3) 2D reconstructions obtained from CT scans of dry skulls without a cervical spine (anthropological material). Our primary goal was to assess X-rays reconstructed from tomographies as an alternative for conventional cephalometric X-rays. Our secondary goal was to assess whether computerized cephalometric analysis was a reproducible technique, both on clinical and anthropological material.

We included 10 consecutive adult patients admitted for orthodontic assessment who had benefited from both lateral cephalometric X-rays and from CT-scans. We also included 10 CT-scans from adult dry skulls without a cervical spine, from the collections of the Muséum National d'Histoire Naturelle (Musée de l'Homme, Paris, France). Cephalometric X-rays were reconstructed from CT-scans of both patients and dry skulls using Carestream® (Carestream Health inc., New York, USA). Simplified Delaire analyses (Top12) were conducted by 2 observers using the Delaire Evolution software (JDel, Nantes, France), on the 3 sets of cephalometric X-rays: (1) 10 standard cephalometric X-rays from the orthodontic patients, (2) 10 reconstructed cephalometric X-rays from the CT scans of the orthodontic patients and (3) 10 reconstructed cephalometric X-rays from the CT scans of the dry skulls. A standard statistical assessment of reproducibility was conducted using correlation coefficients.

We found good inter- and intra-observer reproducibility for standard cephalometric X-rays and reconstructed cephalometric X-rays (Intraclass Correlation Coeficient > 0.75). We did not find any difference for angle measures between the standard cephalometric X-rays and the reconstructed cephalometric X-rays for the group of orthodontic patients (P > 0.05). Delaire cephalometric analysis was not reproducible for cephalometric X-rays without a cervical spine, that is for cephalometric X-rays reconstructed from the CT-scans of dry skulls.

Delaire computerized simplified Top12 analysis was reproducible for lateral cephalometric X-rays and for reconstructions obtained from CT scans with similar angle measures for a given patient. This analysis does not seem to be reliable for dry skulls without upper cervical spine. This study uses the example of one particular computerized cephalometric analysis in order to show that cephalometric analyses can be performed on lateral reconstructed cephalometric X-rays obtained from CT scans.