Premium
TU‐C‐103‐02: Use of New Phantom and Chanelized Hotelling Obeserver (CHO) Based Detectability Algorithm for Demonstrating Variability in Dose Reduction Over CT Scanner Operating Range
Author(s) -
Toth T,
Rohler D,
Izen S,
Maniyedath A,
Kalra M,
Singh S,
Padole A
Publication year - 2013
Publication title -
medical physics
Language(s) - English
Resource type - Journals
SCImago Journal Rank - 1.473
H-Index - 180
eISSN - 2473-4209
pISSN - 0094-2405
DOI - 10.1118/1.4815391
Subject(s) - scanner , imaging phantom , algorithm , nuclear medicine , iterative reconstruction , reduction (mathematics) , mathematics , computer science , physics , optics , artificial intelligence , medicine , geometry
Purpose: Our purpose is to assess detectability performance between filtered backprojection, adaptive iterative and model based iterative reconstruction methods for CT scanners from three different vendors identified as A thru C. The goal is to characterize the variability of detectability performance over the range of radiation dose levels and patient sizes. Methods: We scanned a special phantom containing three sections, having water equivalent diameters (WEQ) of 13.6, 21.2 and 30.9 cm, at six different radiation dose levels over a CTDIvol range of 0.8 to 22 mGy. Each section contained a region with 5 mm (n=3) and 2 mm (n=3) low contrast rods, a uniform region, and a region to measure rod contrasts in the image. Images were obtained using typical body protocols (target helical pitch of 1 and 1 mm slices for each scanner). A Channelized Hotelling Observer (CHO) algorithm with Laguerre‐Gaussian templates was used to evaluate approximately 100,000 images. The collection of detectability results was fitted to a continuous function of dose and patient size. Results: A continuous characterization of the dose reduction percentages over the scanner operating range was obtained for the three scanners. As an example, for scanner A, the dose reduction percentage for the Adaptive Iterative method ranged from 9% to 37%, as compared with FBP. For scanner C, the dose reduction percentage for the Model Based Iterative method ranged from 37% to 66%, as compared with FBP. The continuous detectability characterization captured (1) deviations from quantum noise limited behavior and (2) the presence of artifacts observable at higher flux levels. Conclusion: A rigorous CHO‐based analysis with our phantom can reveal significant performance differences when images are evaluated over a wide dose and size continuum, as compared with evaluating performance at isolated radiation dose and size operating points.