Comparison of STL skull models produced using open-source software versus commercial software

Abstract

Purpose This paper aims to compare the automatic segmentation of medical data and conversion to stereolithography (STL) skull models using open-source software versus commercial software. Design/methodology/approach Both open-source and commercial software used automatic segmentation and post-processing of the data without user intervention, thus avoiding human error. Detailed steps were provided for comparisons and easier to be repeated by other researchers. The results of segmentation, which were converted to STL format were compared using geometric analysis. Findings STL skull models produced using open-source software are comparable with the one produced using commercial software. A comparison of STL skull model produced using InVesalius with STL skull model produced using MIMICS resulted in an average dice similarity coefficient (DSC) of 97.6 ± 0.04 per cent and Hausdorff distance (HD) of 0.01 ± 0.005 mm. Inter-rater study for repeatability on MIMICS software yielded an average DSC of 100 per cent and HD of 0. Social implications The application of open-source software will benefit the small research institutions or hospitals to produce and virtualise three-dimensional model of the skulls for teaching or clinical purposes without having to purchase expensive commercial software. It is also easily reproduceable by other researchers. Originality/value This study is one of the first comparative evaluations of an open-source software with propriety commercial software in producing accurate STL skull models. Inaccurate STL models can lead to inaccurate pre-operative planning or unfit implant.