Quantitative evaluation of noise reduction algorithms for very low dose renal CT perfusion imaging

Xin Liu, Andrew N. Primak, Lifeng Yu, Hua Li, James D. Krier, Lilach O. Lerman, Cynthia H. McCollough

Research output: Chapter in Book/Report/Conference proceedingConference contributionpeer-review

12 Scopus citations


In this paper, we demonstrate a methodology for quantitative evaluation of noise reduction algorithms for very low-dose (1/10 th typical dose) renal CT perfusion imaging. Three types of noise reduction algorithms are evaluated, including the commonly used low pass filtering, edge-preserving algorithms, and spatial-temporal filtering algorithms, such as recently introduced local highly constrained back projection (HYPR-LR) technique and multi-band filtering (MBF). The performance of these noise reduction methods was evaluated in terms of background signal-to-noise ratio (SNR), spatial resolution, fidelity of the time-attenuation curves of renal cortex, and computational speed. The spatial resolution was quantified by an on-scene modulation transfer function (MTF) measurement method. The fidelity of time-attenuation curves was quantified by statistical analysis using a Chi-square test. The results indicate that algorithms employing spatial-temporal correlations of images, such as HYPR and MBF.

Original languageEnglish
Title of host publicationMedical Imaging 2009
Subtitle of host publicationPhysics of Medical Imaging
StatePublished - 2009
EventMedical Imaging 2009: Physics of Medical Imaging - Lake Buena Vista, FL, United States
Duration: Feb 9 2009Feb 12 2009

Publication series

NameProgress in Biomedical Optics and Imaging - Proceedings of SPIE
ISSN (Print)1605-7422


ConferenceMedical Imaging 2009: Physics of Medical Imaging
Country/TerritoryUnited States
CityLake Buena Vista, FL


  • Algorithm
  • Computed tomography
  • Low dose imaging
  • Noise reduction


Dive into the research topics of 'Quantitative evaluation of noise reduction algorithms for very low dose renal CT perfusion imaging'. Together they form a unique fingerprint.

Cite this