A Practical Method to Superimpose Sella Turcica in Serial CBCT Images

Abstract

Introduction: We argue, for studies of growth or treatment changes, superimposed CBCT images ought to be employed routinely. The location of 3D landmarks on curved structures, however, has been complex. In this study, a reliable method of CBCT superimposition was developed with the aid of AMIRA XImage® software. The specific focus was to develop a practical method for cranial base superimpositions at sella turcica in serial CBCT images. Subsequently, evaluations of the anterior curvature of hypophyseal fossa during growth were completed to confirm the stability. Comparisons of surface changes in the superimposed images in different regions were also made. Methods: Samples of pre- and post-orthodontic CBCT data were procured from Case Western Reserve University. Cranial base changes in 10 adolescent patients of Class I and II hypodivergent, normodivergent, and hyperdivergent malocclusions, were observed over an average interval of 2 years and 5 months. Cranial base models were superimposed using a rigid registration technique, and 3D models constructed from manual segmentation in the AMIRA software. Surface distance changes of the entire cranial base were compared to that of the anterior portion of sella turcica. Results: Mean surface distance changes for the entire cranial base and anterior sella were 0.73-1.57 ± 0.71-1.30 mm and 0.37-0.63 ± 0.29-0.43 mm, respectively. The area above a threshold of 0.5 mm was 61.02-74.11% (cranial base) and 40.70-50.94% (anterior sella) as well. Paired t-tests were applied to compare differences between the mean distance and area above the threshold for data of cranial base versus anterior sella. Both illustrated p-values less than 0.0001 and were statistically significant. Intra-rater reliability was evaluated by completing segmentation and registration three consecutive times on each CBCT volume. Mean surface differences were within 0.01 mm with the exception of one patient. In addition, all subjects exhibited less than 1.00% deviation except three patients, which deviated by 1.03%, 1.46%, and 1.64%, respectively. Conclusions: A practical method of superimposition of serial CBCT images was developed. Through production of color maps, the surface distance of the anterior portion of sella turcica was shown to change much less than the clivus and remaining cranial base structures in pre- and post-treatment scans of growing children. Lastly, the method of superimposition developed here may assist future studies of skeletal changes with accuracy.