Speeding Up Colon CAD using KD-tree

Abstract

This paper proposes an efficient technique to speed up our colon CAD software for flat lesion recognition in colon CT images. The proposed technique uses KD-tree to speed up the neighbor searching part in the software. We test our method under Linux system with intel(R) Xeon(R) 2.66 GHz CPU X5355, 12 G memory. Experiments on 50 computed tomographic colonography (CTC) scan database demonstrate the performance of our speeding up method based on KD-tree. It significantly reduces the average processing time from 6,134 s down to 514 s for each CTC scan. © 2014 Springer International Publishing Switzerland.