• Ramandeep Singh, Rajdeep Singh, Britty Baby, and Ashish Suri.
• Department of Neurosurgery, All India Institute of Medical Sciences, New Delhi, India.
• World Neurosurg. 2022 Oct 1; 166: e34e43e34-e43.

BackgroundA variety of applications related to neurosurgical procedures, education, and training require accurate reconstruction of the involved structures from the medical images such as computed tomography (CT). This study evaluates the quality of CT-based reconstruction of dry skull bones for advanced neurosurgical applications. The accuracy and precision of these models were examined with reference optical scanning.MethodsThree consecutive CT and optical scans of different skull bones were acquired and used to develop three-dimensional models. The accuracy of three-dimensional models was examined by manual inspection of the defined anatomical landmarks of the skull. Reproducibility was examined by deviation analysis of the models developed from repeated CT and optical scans.ResultsPrecision was excellent in both the techniques with less than 0.1 mm deviation error. On the interscan evaluation of the CT versus optical scan model, deviations of more than 0.1 mm were observed in 16 out of 21 instances. CT reconstruction using standard segmentation algorithms results in missing bone portions while using the default bone segmentation threshold. The segmentation threshold was varied to construct missing bone regions, and its effect on the iso-surface generation was evaluated. The threshold variation led to increased mean deviations of surfaces up to 0.6 mm.ConclusionsThe study reveals that bone structure, complexity, and segmentation threshold lead to CT reconstruction variability. The trade-off between the desirable model and accepted mean deviation should be considered as per traits of the desired application.Copyright © 2022. Published by Elsevier Inc.

18 keywords...

hide…