skip to main content
OSTI.GOV title logo U.S. Department of Energy
Office of Scientific and Technical Information

Title: SU-F-I-80: Correction for Bias in a Channelized Hotelling Model Observer Caused by Temporally Variable Non-Stationary Noise

Abstract

Purpose: Application of a channelized Hotelling model observer (CHO) over a wide range of x-ray angiography detector target dose (DTD) levels demonstrated substantial bias for conditions yielding low detectability indices (d’), including low DTD and small test objects. The purpose of this work was to develop theory and methods to correct this bias. Methods: A hypothesis was developed wherein the measured detectability index (d’b) for a known test object is positively biased by temporally variable non-stationary noise in the images. Hotelling’s T2 test statistic provided the foundation for a mathematical theory which accounts for independent contributions to the measured d’b value from both the test object (d’o) and non-stationary noise (d’ns). Experimental methods were developed to directly estimate d’o by determining d’ns and subtracting it from d’b, in accordance with the theory. Specifically, d’ns was determined from two sets of images from which the traditional test object was withheld. This method was applied to angiography images with DTD levels in the range 0 to 240 nGy and for disk-shaped iodine-based contrast targets with diameters 0.5 to 4.0 mm. Results: Bias in d’ was evidenced by d’b values which exceeded values expected from a quantum limited imaging system and decreasing objectmore » size and DTD. d’ns increased with decreasing DTD, reaching a maximum of 2.6 for DTD = 0. Bias-corrected d’o estimates demonstrated sub-quantum limited performance of the x-ray angiography for low DTD. Findings demonstrated that the source of non-stationary noise was detector electronic readout noise. Conclusion: Theory and methods to estimate and correct bias in CHO measurements from temporally variable non-stationary noise were presented. The temporal non-stationary noise was shown to be due to electronic readout noise. This method facilitates accurate estimates of d’ values over a large range of object size and detector target dose.« less

Authors:
;  [1]
  1. Mayo Clinic, Rochester, MN (United States)
Publication Date:
OSTI Identifier:
22632140
Resource Type:
Journal Article
Resource Relation:
Journal Name: Medical Physics; Journal Volume: 43; Journal Issue: 6; Other Information: (c) 2016 American Association of Physicists in Medicine; Country of input: International Atomic Energy Agency (IAEA)
Country of Publication:
United States
Language:
English
Subject:
60 APPLIED LIFE SCIENCES; 61 RADIATION PROTECTION AND DOSIMETRY; BIOMEDICAL RADIOGRAPHY; BLOOD VESSELS; CORRECTIONS; IMAGES; IODINE; NOISE; PERFORMANCE; RADIATION DOSES; READOUT SYSTEMS

Citation Formats

Favazza, C, and Fetterly, K. SU-F-I-80: Correction for Bias in a Channelized Hotelling Model Observer Caused by Temporally Variable Non-Stationary Noise. United States: N. p., 2016. Web. doi:10.1118/1.4955908.
Favazza, C, & Fetterly, K. SU-F-I-80: Correction for Bias in a Channelized Hotelling Model Observer Caused by Temporally Variable Non-Stationary Noise. United States. doi:10.1118/1.4955908.
Favazza, C, and Fetterly, K. Wed . "SU-F-I-80: Correction for Bias in a Channelized Hotelling Model Observer Caused by Temporally Variable Non-Stationary Noise". United States. doi:10.1118/1.4955908.
@article{osti_22632140,
title = {SU-F-I-80: Correction for Bias in a Channelized Hotelling Model Observer Caused by Temporally Variable Non-Stationary Noise},
author = {Favazza, C and Fetterly, K},
abstractNote = {Purpose: Application of a channelized Hotelling model observer (CHO) over a wide range of x-ray angiography detector target dose (DTD) levels demonstrated substantial bias for conditions yielding low detectability indices (d’), including low DTD and small test objects. The purpose of this work was to develop theory and methods to correct this bias. Methods: A hypothesis was developed wherein the measured detectability index (d’b) for a known test object is positively biased by temporally variable non-stationary noise in the images. Hotelling’s T2 test statistic provided the foundation for a mathematical theory which accounts for independent contributions to the measured d’b value from both the test object (d’o) and non-stationary noise (d’ns). Experimental methods were developed to directly estimate d’o by determining d’ns and subtracting it from d’b, in accordance with the theory. Specifically, d’ns was determined from two sets of images from which the traditional test object was withheld. This method was applied to angiography images with DTD levels in the range 0 to 240 nGy and for disk-shaped iodine-based contrast targets with diameters 0.5 to 4.0 mm. Results: Bias in d’ was evidenced by d’b values which exceeded values expected from a quantum limited imaging system and decreasing object size and DTD. d’ns increased with decreasing DTD, reaching a maximum of 2.6 for DTD = 0. Bias-corrected d’o estimates demonstrated sub-quantum limited performance of the x-ray angiography for low DTD. Findings demonstrated that the source of non-stationary noise was detector electronic readout noise. Conclusion: Theory and methods to estimate and correct bias in CHO measurements from temporally variable non-stationary noise were presented. The temporal non-stationary noise was shown to be due to electronic readout noise. This method facilitates accurate estimates of d’ values over a large range of object size and detector target dose.},
doi = {10.1118/1.4955908},
journal = {Medical Physics},
number = 6,
volume = 43,
place = {United States},
year = {Wed Jun 15 00:00:00 EDT 2016},
month = {Wed Jun 15 00:00:00 EDT 2016}
}