Publication detail

Preprocessing for Quantitative Statistical Noise Analysis of MDCT Brain Images Reconstructed Using Hybrid Iterative (iDose) Algorithm

WALEK, P. JAN, J. OUŘEDNÍČEK, P. SKOTÁKOVÁ, J. JÍRA, I.

Original Title

Preprocessing for Quantitative Statistical Noise Analysis of MDCT Brain Images Reconstructed Using Hybrid Iterative (iDose) Algorithm

Czech Title

Předzpracování pro kvantitativní statistickou analýzu MDCTobrazů mozků rekonstruovaných hybridní iterativní (iDose) rekonstrukcí

English Title

Preprocessing for Quantitative Statistical Noise Analysis of MDCT Brain Images Reconstructed Using Hybrid Iterative (iDose) Algorithm

Type

journal article

Language

en

Original Abstract

Radiation dose reduction is a very actual problem in medical X-ray CT imaging and plenty of strategies have been introduced recently. Hybrid iterative reconstruction algorithms are one of them enabling dose reduction up to 70 %. Paper describes data preprocessing and feature extraction from iteratively reconstructed images in order to assess their quality in terms of image noise and compare with quality of images reconstructed by conventional filtered back projection. Preprocessing stage is consisted from correction of the stair-step artifact and fast, precise bones and soft tissue segmentation. Noise patterns of differently reconstructed images can therefore be examined separately in these tissue types. In order to remove anatomical structures and obtain pure noise subtraction of images reconstructed by iDose from images reconstructed by filtered back projection is performed. Results of these subtractions are called residual noise images and are used to further extraction of noise parameters. The noise parameters which are intended to serve as input data for further multidimensional statistical analysis are standard deviation and noise power spectrum of residual noise. Performed approach enables evaluation of noise properties in whole volume of real patient data in contrast with noise analysis performed in small region of interest or in images of phantoms.

Czech abstract

Redukce dávky v zobrazení rentgenovou výpočetní tomografií je velmi aktuální problém a v nedávné době bylo představeno několik nových přístupů k jeho řešení. Jedním z těchto přístupů je zavedení hybridních iterativních rekonstrukcí, které dovolují redukci dávky až o 70%. Příspěvek popisuje předzpracování a extrakci parametrů šumu z iterativně rekonstruovaných obrazů. Tyto parametry budou následně použity k porovnání šumových vlastností iterativně rekonstruovaných obrazů a obrazů rekonstruovaných pomocí standardně používané filtrované zpětné projekce. Fáze předzpracování obrazů je složena ze dvou částí a to odstranění schodovitého artefaktu a segmentace kostí a měkkých tkání. Segmentace je provedena z důvodu porovnání šumových vlastností různých tkání. Pomocí subtrakce různě rekonstruovaných obrazů jsou odstraněny anatomické struktury a jsou získány obrazy, ve kterých je pouze šum. Tyto obrazy jsou pojmenovány jako obrazy reziduálního šumu a právě z nich jsou extrahovány šumové parametry. Parametry šumu, které budou v dalším výzkumu použity pro vícerozměrnou statistickou analýzu, jsou směrodatná odchylka a výkonové spektrum reziduálního šumu. Výhodou představeného předzpracování je fakt, že je analyzován celý objem reálných pacientských dat na rozdíl od běžně používaných šumových analýz fantomů nebo malých zájmových oblastí pacientských dat.

English abstract

Radiation dose reduction is a very actual problem in medical X-ray CT imaging and plenty of strategies have been introduced recently. Hybrid iterative reconstruction algorithms are one of them enabling dose reduction up to 70 %. Paper describes data preprocessing and feature extraction from iteratively reconstructed images in order to assess their quality in terms of image noise and compare with quality of images reconstructed by conventional filtered back projection. Preprocessing stage is consisted from correction of the stair-step artifact and fast, precise bones and soft tissue segmentation. Noise patterns of differently reconstructed images can therefore be examined separately in these tissue types. In order to remove anatomical structures and obtain pure noise subtraction of images reconstructed by iDose from images reconstructed by filtered back projection is performed. Results of these subtractions are called residual noise images and are used to further extraction of noise parameters. The noise parameters which are intended to serve as input data for further multidimensional statistical analysis are standard deviation and noise power spectrum of residual noise. Performed approach enables evaluation of noise properties in whole volume of real patient data in contrast with noise analysis performed in small region of interest or in images of phantoms.

Keywords

X-ray computed tomography, dose reduction, skull segmentation, noise power spectrum

RIV year

2012

Released

28.06.2012

Pages from

73

Pages to

80

Pages count

8

BibTex


@article{BUT92031,
  author="Petr {Walek} and Jiří {Jan} and Petr {Ouředníček} and Jarmila {Skotáková} and Igor {Jíra}",
  title="Preprocessing for Quantitative Statistical Noise Analysis of MDCT Brain Images Reconstructed Using Hybrid Iterative (iDose) Algorithm",
  annote="Radiation dose reduction is a very actual problem in medical X-ray CT imaging and plenty of strategies have been introduced recently. Hybrid iterative reconstruction algorithms are one of them enabling dose reduction up to 70 %. Paper describes data preprocessing and feature extraction from iteratively reconstructed images in order to assess their quality in terms of image noise and compare with quality of images reconstructed by conventional filtered back projection. Preprocessing stage is consisted from correction of the stair-step artifact and fast, precise bones and soft tissue segmentation. Noise patterns of differently reconstructed images can therefore be examined separately in these tissue types. In order to remove anatomical structures and obtain pure noise subtraction of images reconstructed by iDose from images reconstructed by filtered back projection is performed. Results of these subtractions are called residual noise images and are used to further extraction of noise parameters. The noise parameters which are intended to serve as input data for further multidimensional statistical analysis are standard deviation and noise power spectrum of residual noise. Performed approach enables evaluation of noise properties in whole volume of real patient data in contrast with noise analysis performed in small region of interest or in images of phantoms.",
  chapter="92031",
  howpublished="print",
  number="1",
  volume="20",
  year="2012",
  month="june",
  pages="73--80",
  type="journal article"
}