Comparison between human and model observer performance in low-contrast detection tasks in CT images: application to images reconstructed with filtered back projection and iterative algorithms

Abstract

RMS, Veldkamp WJH. Comparison between human and model observer performance in lowcontrast detection tasks in CT images: application to images reconstructed with filtered back projection and iterative algorithms. Br J Radiol 2014;87:20140014. FULL PAPERComparison between human and model observer performance in low-contrast detection tasks in CT images: application to images reconstructed with filtered back projection and iterative algorithms Objective: To compare low-contrast detectability (LCDet) performance between a model [non-pre-whitening matched filter with an eye filter (NPWE)] and human observers in CT images reconstructed with filtered back projection (FBP) and iterative [adaptive iterative dose reduction three-dimensional (AIDR 3D; Toshiba Medical Systems, Zoetermeer, Netherlands)] algorithms. Methods: Images of the Catphan® phantom (Phantom Laboratories, New York, NY) were acquired with Aquilion ONE™ 320-detector row CT (Toshiba Medical Systems, Tokyo, Japan) at five tube current levels (20-500 mA range) and reconstructed with FBP and AIDR 3D. Samples containing either low-contrast objects (diameters, 2-15 mm) or background were extracted and analysed by the NPWE model and four human observers in a two-alternative forced choice detection task study. Proportion correct (PC) values were obtained for each analysed object and used to compare human and model observer performances. An efficiency factor (h) was calculated to normalize NPWE to human results. Results: Human and NPWE model PC values (normalized by the efficiency, h 5 0.44) were highly correlated for the whole dose range. The Pearson's product-moment correlation coefficients (95% confidence interval) between human and NPWE were 0.984 (0.972-0.991) for AIDR 3D and 0.984 (0.971-0.991) for FBP, respectively. Bland-Altman plots based on PC results showed excellent agreement between human and NPWE [mean absolute difference 0.5 6 0.4%; range of differences (24.7%, 5.6%)]. Conclusion: The NPWE model observer can predict human performance in LCDet tasks in phantom CT images reconstructed with FBP and AIDR 3D algorithms at different dose levels. Advances in knowledge: Quantitative assessment of LCDet in CT can accurately be performed using software based on a model observer.

Development of a method to create uniform phantoms for task‐based assessment of CT image quality

“…Several LCD phantoms are commercially available and were previously used for task‐based image quality assessment 2–4 . The ACR accreditation phantom (Gammex, Middleton, WI) contains cylinders of 2 to 25 mm diameter and 6 HU contrast.…”

Construction of realistic phantoms from patient images and a commercial three-dimensional printer

“…[14][15][16][17][18][19][20] Several previous studies have demonstrated a good correlation between the performance of human observer and model observer. [17][18][19][21][22][23] Most of these studies were performed using phantoms with uniform background due to simplicity and availability of these phantoms. However, to assess image quality and radiation dose reduction in clinical CT imaging, physical phantoms having realistic background textures and lesions are highly desirable as the performance of reconstruction algorithms may be affected by anatomic background (i.e., structured noise).…”

A simple method for low-contrast detectability, image quality and dose optimisation with CT iterative reconstruction algorithms and model observers

“…Four-alternative forced-choice (4-AFC) [ 13 ] test was executed to evaluate low-contrast spatial resolution by five radiologists with at least 15 years of experience in clinical CT and four experienced radiology technicians [ 14 ]. Observers were trained on all technical aspects and objectives of the study and frontal training was performed through examples before the test.…”