TY - JOUR
T1 - Image noise, cnr, and Detectability of low-contrast, low-attenuation liver lesions in a Phantom
T2 - Effects of Radiation Exposure, Phantom Size, Integrated Circuit Detector, and Iterative Reconstruction1
AU - Goenka, Ajit H.
AU - Herts, Brian R.
AU - Dong, Frank
AU - Obuchowski, Nancy A.
AU - Primak, Andrew N.
AU - Karim, Wadih
AU - Baker, Mark E.
N1 - Publisher Copyright:
© 2016 RSNA.
PY - 2016/8
Y1 - 2016/8
N2 - Purpose: To assess image noise, contrast-to-noise ratio (CNR) and detectability of low-contrast, low-attenuation liver lesions in a semianthropomorphic phantom by using either a discrete circuit (DC) detector and filtered back projection (FBP) or an integrated circuit (IC) detector and iterative reconstruction (IR) with changes in radiation exposure and phantom size. Materials and Methods: An anthropomorphic phantom without or with a 5-cm-thick fat-mimicking ring (widths, 30 and 40 cm) containing liver inserts with four spherical lesions was scanned with five exposure settings on each of two computed tomography scanners, one equipped with a DC detector and the other with an IC detector. Images from the DC and IC detector scanners were reconstructed with FBP and IR, respectively. Image noise and lesion CNR were measured. Four radiologists evaluated lesion presence on a five-point diagnostic confidence scale. Data analyses included receiver operating characteristic (ROC) curve analysis and noninferiority analysis. Results: The combination of IC and IR significantly reduced image noise (P <.001) (with the greatest reduction in the 40-cm phantom and at lower exposures) and improved lesion CNR (P <.001). There was no significant difference in area under the ROC curve between detector-reconstruction combinations at fixed exposure for either phantom. Reader accuracy with IC-IR was noninferior at 50% (100 mAs [effective]) and 25% (300 mAs [effective]) exposure reduction for the 30-and 40-cm phantoms, respectively (adjusted P <.001 and.04 respectively). IC-IR improved readers' confidence in the presence of a lesion (P =.029) independent of phantom size or exposure level. Conclusion: IC-IR improved objective image quality and lesion detection confidence but did not result in superior diagnostic accuracy when compared with DC-FBP. Moderate exposure reductions maintained comparable diagnostic accuracy for both detectorreconstruction combinations. Lesion detection in the 40-cm phantom was inferior at smaller exposure reduction than in the 30-cm phantom.
AB - Purpose: To assess image noise, contrast-to-noise ratio (CNR) and detectability of low-contrast, low-attenuation liver lesions in a semianthropomorphic phantom by using either a discrete circuit (DC) detector and filtered back projection (FBP) or an integrated circuit (IC) detector and iterative reconstruction (IR) with changes in radiation exposure and phantom size. Materials and Methods: An anthropomorphic phantom without or with a 5-cm-thick fat-mimicking ring (widths, 30 and 40 cm) containing liver inserts with four spherical lesions was scanned with five exposure settings on each of two computed tomography scanners, one equipped with a DC detector and the other with an IC detector. Images from the DC and IC detector scanners were reconstructed with FBP and IR, respectively. Image noise and lesion CNR were measured. Four radiologists evaluated lesion presence on a five-point diagnostic confidence scale. Data analyses included receiver operating characteristic (ROC) curve analysis and noninferiority analysis. Results: The combination of IC and IR significantly reduced image noise (P <.001) (with the greatest reduction in the 40-cm phantom and at lower exposures) and improved lesion CNR (P <.001). There was no significant difference in area under the ROC curve between detector-reconstruction combinations at fixed exposure for either phantom. Reader accuracy with IC-IR was noninferior at 50% (100 mAs [effective]) and 25% (300 mAs [effective]) exposure reduction for the 30-and 40-cm phantoms, respectively (adjusted P <.001 and.04 respectively). IC-IR improved readers' confidence in the presence of a lesion (P =.029) independent of phantom size or exposure level. Conclusion: IC-IR improved objective image quality and lesion detection confidence but did not result in superior diagnostic accuracy when compared with DC-FBP. Moderate exposure reductions maintained comparable diagnostic accuracy for both detectorreconstruction combinations. Lesion detection in the 40-cm phantom was inferior at smaller exposure reduction than in the 30-cm phantom.
UR - http://www.scopus.com/inward/record.url?scp=84979896136&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=84979896136&partnerID=8YFLogxK
U2 - 10.1148/radiol.2016151621
DO - 10.1148/radiol.2016151621
M3 - Article
C2 - 26937709
AN - SCOPUS:84979896136
SN - 0033-8419
VL - 280
SP - 475
EP - 482
JO - Radiology
JF - Radiology
IS - 2
ER -