Zhang Y.,Computer Science and Technology Academy of NWU |
Feng J.,Computer Science and Technology Academy of NWU |
Cui L.,Computer Science and Technology Academy of NWU |
Zhao Y.-K.,Computer Science and Technology Academy of NWU |
He X.-W.,Computer Science and Technology Academy of NWU
Xitong Fangzhen Xuebao / Journal of System Simulation | Year: 2014
Centerline extraction of colon lumen is one of the important procedures for computer aid detection of colorectal diseases. In the mean time, the evaluation of extracted centerlines is still an open problem. Based on centerline and profile parameter functions, 20 analytic geometric models with centerline standard were generated, which simulated the colon structure characteristics, i. e. tubular shape, ellipse lumen and high curvature. Furthermore, Dijkstra's shortest path algorithm (DIJ), Maximal Spanning Tree algorithm (MST) and the proposed fast centerline extraction algorithm based on Maximal Spanning Tree (FMST) were compared based on simulated models and a set of proposed quantitative criteria for evaluation. The experimental results demonstrate that MST and FMST achieve better accurate rates than DIJ, while avoiding hugging the corners around sharp turns. The proposed FMST is the best extraction algorithm which can speed up MST 80%. © Journal of System Simulation.