文献类型： 会议论文

第一作者： Daniel Jie Yuan Chin

作者： Daniel Jie Yuan Chin ¹ ; Ahmad Sufril Azlan Mohamed ¹ ; Khairul Anuar Shariff ² ; Kunio Ishikawa ³ ;

作者机构： 1.School of Computer Sciences, Universiti Sains Malaysia

2.School of Materials and Mineral Resources Engineering, Universiti Sains Malaysia

3.Department of Biomaterials, Faculty of Dental Science, Kyushu University

关键词： Fast 3D reconstruction;Enhanced marching cubes 33;Large CT images

会议名称： International Visual Informatics Conference

主办单位：

页码： 374-384

摘要： With the advancement in three-dimensional technologies, three-dimensional reconstruction of medical images serves as reliable assistance for doctors and surgeons in evaluating and diagnosing bone defects. Amongst the existing reconstruction methods, the Marching Cubes algorithm is highly popular in the surface rendering research study. There are many improvements made over the Marching Cubes algorithm, but due to the relatively small image datasets used during evaluation, it is difficult to judge the effectiveness of the improvements on large image datasets and the reconstructed models may not be viewable in lower-end specs digital devices like tablets and smartphones. Thus, an enhancement over the extended Marching Cubes 33 with graphics processing unit acceleration to improve the reconstruction accuracy, execution time, and model portability for large image datasets is proposed in this study. The obtained results show that the proposed enhancement successfully increased the accuracy by 5.29%, decreased the execution time by 11.16%, and decreased the number of vertices and faces by 73.72%. This shows that it is possible to view bone defect models with a high similarity percentage on lower-end spec digital devices and print them out with a three-dimensional printer.

分类号： G202-53