Fast image reconstruction with cone-beam ART algorithm

Shunli Zhang, Dinghua Zhang, Kuidong Huang, Yunyong Cheng

Research output: Contribution to journalArticlepeer-review

8 Scopus citations

Abstract

To improve the reconstruction speed of cone-beam ART, this paper presents a fast algorithm based on projection for voxel traversal and intersection calculation along 3D ray. Firstly, the 3D ray is projected onto two perpendicular planes; then, the intersection situation between the projection line and planes is calculated; finally, the voxel index and length traversed by the 3D ray can be determined from the above intersection situation. Using this algorithm to calculate the weight coefficients during reconstructing on real-time can not only save large amount of memory, but also improve the speed of projection and back-projection. Based on the characteristic of the proposed algorithm, a strategy of processing column data first is used during reconstructing, thus unnecessary computation is reduced and the reconstruction speed is improved greatly. The result shows that the algorithm is very effective and the reconstruction speed is improved about 17 times compared with the conventional Siddon algorithm.

Original languageEnglish
Pages (from-to)887-892
Number of pages6
JournalYi Qi Yi Biao Xue Bao/Chinese Journal of Scientific Instrument
Volume30
Issue number4
StatePublished - Apr 2009

Keywords

  • Algebraic reconstruction technique
  • Cone-beam CT
  • Image reconstruction
  • Voxel traversal

Fingerprint

Dive into the research topics of 'Fast image reconstruction with cone-beam ART algorithm'. Together they form a unique fingerprint.

Cite this