Prior Information Enhanced Adversarial Learning for kVp Switching CT

EasyChair Preprint 11239

Abstract

Dual energy computed tomography (DECT) can provide both structural and material information of the scanned object, and has been widely used in the medical field. However, patients may suffer from genetic damage and cancer under long-term high radiation dose of x-ray exposure. To reduce radiation dose and ensure optimal hardware cost. This work studies the switching technology based on the x-ray tube voltage (kVp), which only requires the traditional energy integration detector and the ray source. However, the kVp switching technology faces the problems of low sampling rate of each energy spectrum and the spatial misalignment of projection data of different energy spectrum. Thus, this study introduces an adversarial learning mechanism and proposes a Prior Information enhanced Projection data Inpainting Network (PINet). The experimental results show that the PINet framework is a promising approach for sparse-view angle DECT imaging.

Keyphrases: Adversarial Learning, adversarial learning., dual energy computed tomography, kVp switching technology, projection data inpainting

@booklet{EasyChair:11239,
  author    = {Yizhong Wang and Ailong Cai and Ningning Liang and Shaoyu Wang and Junru Ren and Xinrui Zhang and Lei Li and Bin Yan},
  title     = {Prior Information Enhanced Adversarial Learning for kVp Switching CT},
  howpublished = {EasyChair Preprint 11239},
  year      = {EasyChair, 2023}}