A novel image quality assessment index for edge-aware noise reduction in low-dose fluoroscopy: Preliminary results

Emilio Andreozzi, Maria Agnese Pirozzi, Antonio Fratini, Giuseppe Cesarelli, Mario Cesarelli, Paolo Bifulco

Research output: Chapter in Book/Report/Conference proceedingConference publication

Abstract

X-ray fluoroscopy is a medical imaging modality that provides continuous real-time screening of patient’s organs and various radiopaque surgical objects. Fluoroscopy usually requires long and unpredictable exposure times, thus radiation intensity must be heavily reduced to limit patient’s dose. This gives rise to the well-known Poisson noise, which results in very poor image quality. Commercial fluoroscopes usually improve image quality via real-time temporal averaging, which produces motion blur in moving scenes. The Noise Variance Conditioned Average (NVCA) algorithm exploits the a priori knowledge of Poisson noise statistics to provide efficient noise reduction, while preserving the edges of moving objects. However, accurate setting of NVCA parameters is required to achieve the best results, and this could be supported by image quality assessment (IQA) indices. This study presents a novel, edge-aware IQA index, named Sensitivity of Edge Detection (SED), and compares it against the well-established Feature Similarity (FSIM) index, to assess their efficiency in determining the optimal parameters for NVCA. The preliminary results obtained in this study suggest SED could be more efficient than FSIM in identifying the best trade-off between noise reduction and edge preservation, and could be also used to determine the optimal parameters of other denoising algorithms.

Original languageEnglish
Title of host publication2020 8th E-Health and Bioengineering Conference, EHB 2020
PublisherIEEE
ISBN (Electronic)9781728188034
DOIs
Publication statusPublished - 10 Dec 2020
Event8th E-Health and Bioengineering Conference, EHB 2020 - Virtual, Iasi, Romania
Duration: 29 Oct 202030 Oct 2020

Publication series

Name2020 8th E-Health and Bioengineering Conference, EHB 2020
PublisherIEEE
ISSN (Print)2575-5137
ISSN (Electronic)2575-5145

Conference

Conference8th E-Health and Bioengineering Conference, EHB 2020
CountryRomania
CityVirtual, Iasi
Period29/10/2030/10/20

Keywords

  • Edge-aware
  • Fluoroscopy
  • Image quality assessment
  • Poisson denoising
  • X-ray imaging

Fingerprint Dive into the research topics of 'A novel image quality assessment index for edge-aware noise reduction in low-dose fluoroscopy: Preliminary results'. Together they form a unique fingerprint.

Cite this