Real-time quasi-3D tomographic reconstruction

Jan‐Willem Buurlage, Holger Kohr, Willem Jan Palenstijn, Kees Joost BatenburgMeasurement 2018 · DOI

Abstract

Current approaches to tackle the computational challenges in real-time tomographic reconstruction can be roughly subdivided into two groups. First, reconstruction algorithms that are computationally more efficient are being adopted. Two examples of this are the gridrec method [6, 7] and methods based on the log-polar Radon transform Second, reconstruction algorithms can be run in parallel, either on distributed computer clusters or specialized hardware such as GPUs However, while these approaches can lead to a dramatic reduction in reconstruction times, the computational demands for reconstructing the full 3D volume remain a bottleneck for truly real-time tomographic reconstruction. By realizing that while currently often full 3D reconstructions are made, the reconstructed volume is primarily viewed slice by slice, we observe that more computational work is done than necessary.
BibTeX
@inproceedings{buurlage2018real,
  author = {Jan‐Willem Buurlage and Holger Kohr and Willem Jan Palenstijn and Kees Joost Batenburg},
  title = {{Real-time quasi-3D tomographic reconstruction}},
  booktitle = {Measurement},
  year = {2018},
  doi = {10.1088/1361-6501/aab754},
}
Figure