Raw-data-based iterative reconstruction versus filtered back projection: image quality of low-dose chest computed...

The objective was to compare standard-dose chest computed tomography (CT) reconstructed with filtered back projection (FBP) versus low-dose images with FBP and raw-data-based iterative reconstruction.

Eighty-seven consecutive patients (46 male; mean age, 54.54 ± 16.12; mean body mass index, 24.58 ± 4.07) referred for initial chest CT with full-dose examinations [mean dose-length product (DLP), 183.37 ± 44.13 mGy · cm] and follow-up chest CT with half-dose examinations (mean DLP, 91.08 ± 23.81 mGy · cm) were included. The full-dose protocol was reconstructed with FBP; the half-dose protocol was reconstructed with FBP and sinogram-affirmed iterative reconstruction (SAFIRE). Noise and signal-to-noise ratio were compared using a paired Student's t test; subjective image quality and lesion conspicuity were compared using Wilcoxon signed ranks test.

Actual radiation dose of follow-up CT was about 50% (49.26% ± 2.62%) of standard-dose protocol. Compared to full-dose images with FBP, there was no significant difference in half-dose images with SAFIRE in the objective noise (ascending aorta: P=.38, descending aorta: P=.70, trachea on mediastinal images: P=.37) and SNR (ascending aorta: P=.14, descending aorta: P=.72, trachea on mediastinal images: P=.06) on mediastinal images. Noise was significantly lower (P<.001) and SNR was significantly higher (P<.001) in half-dose images with SAFIRE on lung images. Noise was significantly higher (P<.001) and SNR was significantly lower (P<.001) in half-dose images with FBP. Subjective image quality was similar on both mediastinal images (P=.317) and lung images (P=.614) of half-dose SAFIRE images versus full-dose FBP images. Lesion conspicuity was also similar. Subjective image quality was significantly lower on both mediastinal images (P<.001) and lung images (P<.001) of half-dose FBP images versus full-dose FBP images. The conspicuity of some lesions was significantly lower (ground-glass opacity, P<.0001; ill-defined micronodule, P<.0001; lung cyst, P<.0001; emphysematous lesion, P=.003) on half-dose FBP versus full-dose FBP images.

Compared to full-dose CT images reconstructed with the conventional FBP algorithm, SAFIRE with three iterations could provide similar or better image quality at 50% less dose.