Premium
TH‐E‐110‐11: The Effect of Region‐Of‐Interest Z‐Axis Collimation on the Image Quality of a Commercial Cone Beam CT Imaging System
Author(s) -
Goodsitt M,
Larson S,
Christodoulou E
Publication year - 2011
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 - 10.1118/1.3613595
Subject(s) - collimated light , imaging phantom , optics , cone beam computed tomography , image quality , physics , materials science , nuclear medicine , image resolution , computed tomography , medicine , laser , image (mathematics) , radiology , computer science , artificial intelligence
Purpose: A study was performed to determine the degree to which region‐ of‐interest z‐axis collimation can improve the image quality of a cone‐beam CT system. Methods: The linearity and low and high‐contrast resolution modules of a 20‐cm diameter Catphan 600 phantom were imaged with a Siemens Zeego Angiography system using three “DynaCT” cone beam CT modes. These included (a) DR‐H Head (20‐sec), (b) DSA‐H (8‐sec), and (c) DR‐Body (8‐sec). All modes employ automatic exposure control (AEC), which can vary the kVp, mA and pulse width throughout the scan. Image quality was compared in 5‐mm thick slices obtained with full‐field (19‐cm) collimation and minimal‐width (2.3‐cm) collimation. Body scans (c) were compared with and without saline bags added at the periphery of the Catphan to create a larger phantom. Results: The contrast‐to‐noise‐ratios (CNRs) of the 15‐mm diameter, nominal 1% (10HU) disk in the Catphan when imaged with full‐field collimation were 1.28, 1.93, and 1.23 for DynaCT modes a, b, and c, respectively. With 2.3‐cm collimation, the corresponding CNRˈs improved to 2.13, 2.46, and 1.45. The CNRˈs for images of the large phantom degraded to 0.51 for full‐field and 0.55 for 2.3‐ cm collimation. Plots of measured CT numbers of plastics in the linearity section as a function of mass density were linear (R‐squared >0.99) in all cases, with slopes that varied from 761 to 1006 HU/(g/cc) depending upon collimation, phantom size and the AEC selected kVp. Spatial resolution was 12 lp/cm for mode (a), and it was 8–;9 lp/cm for all other modes including scans with the large phantom. Conclusions: Collimation significantly improves CNR for a small patient/phantom, but is less effective for large patients/phantoms. Collimation and phantom size have minimal effect on spatial resolution. They affect the slope and intercept but not the degree of linearity of CT number with density.