TY - JOUR
T1 - Lung nodule detection performance in five observers on computed tomography (CT) with adaptive iterative dose reduction using three-dimensional processing (AIDR 3D) in a Japanese multicenter study
T2 - Comparison between ultra-low-dose CT and low-dose CT by receiver-operating characteristic analysis
AU - Nagatani, Yukihiro
AU - Takahashi, Masashi
AU - Murata, Kiyoshi
AU - Ikeda, Mitsuru
AU - Yamashiro, Tsuneo
AU - Miyara, Tetsuhiro
AU - Koyama, Hisanobu
AU - Koyama, Mitsuhiro
AU - Sato, Yukihisa
AU - Moriya, Hiroshi
AU - Noma, Satoshi
AU - Tomiyama, Noriyuki
AU - Ohno, Yoshiharu
AU - Murayama, Sadayuki
N1 - Publisher Copyright:
© 2015 The Authors.
PY - 2015/7/1
Y1 - 2015/7/1
N2 - Abstract Purpose To compare lung nodule detection performance (LNDP) in computed tomography (CT) with adaptive iterative dose reduction using three dimensional processing (AIDR3D) between ultra-low dose CT (ULDCT) and low dose CT (LDCT). Materials and methods This was part of the Area-detector Computed Tomography for the Investigation of Thoracic Diseases (ACTIve) Study, a multicenter research project being conducted in Japan. Institutional Review Board approved this study and informed consent was obtained. Eighty-three subjects (body mass index, 23.3 ± 3.2) underwent chest CT at 6 institutions using identical scanners and protocols. In a single visit, each subject was scanned using different tube currents: 240, 120 and 20 mA (3.52, 1.74 and 0.29 mSv, respectively). Axial CT images with 2-mm thickness/increment were reconstructed using AIDR3D. Standard of reference (SOR) was determined based on CT images at 240 mA by consensus reading of 2 board-certificated radiologists as to the presence of lung nodules with the longest diameter (LD) of more than 3 mm. Another 5 radiologists independently assessed and recorded presence/absence of lung nodules and their locations by continuously-distributed rating in CT images at 20 mA (ULDCT) and 120 mA (LDCT). Receiver-operating characteristic (ROC) analysis was used to evaluate LNDP of both methods in total and also in subgroups classified by LD (>4, 6 and 8 mm) and nodular characteristics (solid and ground glass nodules). Results For SOR, 161 solid and 60 ground glass nodules were identified. No significant difference in LNDP for entire solid nodules was demonstrated between both methods, as area under ROC curve (AUC) was 0.844 ± 0.017 in ULDCT and 0.876 ± 0.026 in LDCT (p = 0.057). For ground glass nodules with LD 8 mm or more, LNDP was similar between both methods, as AUC 0.899 ± 0.038 in ULDCT and 0.941 ± 0.030 in LDCT. (p = 0.144). Conclusion ULDCT using AIDR3D with an equivalent radiation dose to chest x-ray could have comparable LNDP to LDCT with AIDR3D except for smaller ground glass nodules in cases with normal range body habitus.
AB - Abstract Purpose To compare lung nodule detection performance (LNDP) in computed tomography (CT) with adaptive iterative dose reduction using three dimensional processing (AIDR3D) between ultra-low dose CT (ULDCT) and low dose CT (LDCT). Materials and methods This was part of the Area-detector Computed Tomography for the Investigation of Thoracic Diseases (ACTIve) Study, a multicenter research project being conducted in Japan. Institutional Review Board approved this study and informed consent was obtained. Eighty-three subjects (body mass index, 23.3 ± 3.2) underwent chest CT at 6 institutions using identical scanners and protocols. In a single visit, each subject was scanned using different tube currents: 240, 120 and 20 mA (3.52, 1.74 and 0.29 mSv, respectively). Axial CT images with 2-mm thickness/increment were reconstructed using AIDR3D. Standard of reference (SOR) was determined based on CT images at 240 mA by consensus reading of 2 board-certificated radiologists as to the presence of lung nodules with the longest diameter (LD) of more than 3 mm. Another 5 radiologists independently assessed and recorded presence/absence of lung nodules and their locations by continuously-distributed rating in CT images at 20 mA (ULDCT) and 120 mA (LDCT). Receiver-operating characteristic (ROC) analysis was used to evaluate LNDP of both methods in total and also in subgroups classified by LD (>4, 6 and 8 mm) and nodular characteristics (solid and ground glass nodules). Results For SOR, 161 solid and 60 ground glass nodules were identified. No significant difference in LNDP for entire solid nodules was demonstrated between both methods, as area under ROC curve (AUC) was 0.844 ± 0.017 in ULDCT and 0.876 ± 0.026 in LDCT (p = 0.057). For ground glass nodules with LD 8 mm or more, LNDP was similar between both methods, as AUC 0.899 ± 0.038 in ULDCT and 0.941 ± 0.030 in LDCT. (p = 0.144). Conclusion ULDCT using AIDR3D with an equivalent radiation dose to chest x-ray could have comparable LNDP to LDCT with AIDR3D except for smaller ground glass nodules in cases with normal range body habitus.
UR - http://www.scopus.com/inward/record.url?scp=84930182326&partnerID=8YFLogxK
UR - http://www.scopus.com/inward/citedby.url?scp=84930182326&partnerID=8YFLogxK
U2 - 10.1016/j.ejrad.2015.03.012
DO - 10.1016/j.ejrad.2015.03.012
M3 - Article
C2 - 25892051
AN - SCOPUS:84930182326
SN - 0720-048X
VL - 84
SP - 1401
EP - 1412
JO - European journal of radiology
JF - European journal of radiology
IS - 7
M1 - 7070
ER -