A step-by-step method for producing 3D crania models from CT data

Abstract

Abstract Objectives Modern computed tomography (CT) databases offer a valuable resource for obtaining skeletal reconstructions and contemporary population data. However, researchers may not utilise CT data due to limited funds for proprietary modelling software, or from a lack of awareness of visualisation techniques. This paper presents a step-by-step method for creating accurate 3D crania models from CT data using the free and open-source platform 3D Slicer. This method is tested to (1) establish if novice users can produce 3D crania models following the steps, and (2) determine if these models are accurate to models from experienced users. Materials and methods The step-by-step method was recorded and tested by five observers who each produced twenty 3D models using clinical sinus CT scans (n = 20). The models (n = 100) were evaluated through a quantitative mesh comparison to establish the accuracy with experienced users and against novice users. Results The mesh comparison between the models from the experienced observers resulted in an average absolute mean distance of 0.4 mm, with 99% of models accurate to within 0.5 mm. The novice observers were able to create robust 3D models following the step-by-step method with average absolute mean distances of 0.5 to 0.6 mm, and 95% of the mean distances within 1 mm of the reference model. Conclusion All of the crania models produced were comparably accurate with minor variances seen in the background noise and orbital bone modelling. The tested method is accessible and suitable for use with modern CT databases and for forensic reconstruction approaches.