Abstract Archives of the RSNA, 2011
LL-PHS-WE6A
Quantitative Evaluation of Image Quality in Computed Tomography: A Comparison of Iterative Reconstruction and Filter
Scientific Informal (Poster) Presentations
Presented on November 30, 2011
Presented as part of LL-PHS-WE: Physics
Raoul M. S. Joemai, Abstract Co-Author: Research grant, Toshiba Corporation
Wouter J. H. Veldkamp PhD, Presenter: Nothing to Disclose
Jacob Geleijns PhD, Abstract Co-Author: Nothing to Disclose
The aim of this study was to develop quantitative techniques for evaluation of image quality in computed tomography and to apply these techniques to iterative reconstructions and filtered backprojection (FBP).
Three image quality measures were assessed, i.e. spatial resolution (full width at half maximum, FWHM), noise power spectrum (NPS), and low contrast detectability (LCD). The CATPHAN phantom (The Phantom Laboratory, NY, USA) was scanned with a Toshiba Aquilion ONE scanner (Toshiba Medical Systems, volume acquisition, tube voltage 120 kV, tube charge 250-125-65-30-15mAs, beam collimation 320x0.5 mm). Reconstructions were made with FBP and iterative reconstruction (AIDR). Spatial resolution was evaluated as the FWHM of the response of a small tungsten bead with diameter 0.18 mm. Noise was characterized through estimation of the NPS from CT images of a homogeneous section of the CATPHAN phantom. LCD was assessed for the low contrast inserts in the phantom using a human observer model.
FWHM was in axial plane and z-plane similar for AIDR and FBP. Differences for the shape of the NPS were not substantial. At all frequencies AIDR showed a significant decrease of the noise power compared to FBP. LCD was better in AIDR compared to FBP at equal dose levels; LCD in AIDR images at 30mAs was similar compared to FBP images at 65mAs but such relative improvements were smaller at higher dose levels.
Image quality in AIDR shows advantages compared to FBP with regard to LCD with similar FWHM, especially at low dose levels. Observer studies with clinical images should be performed to confirm the potential dose reduction of AIDR.
AIDR offers better performance compared to FBP with regard to noise and LCD. Assessment of AIDR requires sophisticated methods for quantification of image quality and cannot only be based on SD of HU.
Joemai, R,
Veldkamp, W,
Geleijns, J,
Quantitative Evaluation of Image Quality in Computed Tomography: A Comparison of Iterative Reconstruction and Filter. Radiological Society of North America 2011 Scientific Assembly and Annual Meeting, November 26 - December 2, 2011 ,Chicago IL.
http://archive.rsna.org/2011/11034269.html