Relationships of clinical protocols and reconstruction kernels with image quality and radiation dose in a 128-slice CT scanner: study with an anthropomorphic and water phantom.

PURPOSE: The aim of this study was to explore the relationship of scanning parameters (clinical protocols), reconstruction kernels and slice thickness with image quality and radiation dose in a DSCT.
MATERIALS AND METHODS: The chest of an anthropomorphic phantom was scanned on a DSCT scanner (Siemens Somatom Definition flash) using different clinical protocols, including single- and dual-energy modes. Four scan protocols were investigated: 1) single-source 120kV, 110mAs, 2) single-source 100kV, 180mAs, 3) high-pitch 120kV, 130mAs and 4) dual-energy with 100/Sn140kV, eff.mAs 89, 76. The automatic exposure control was switched off for all the scans and the CTDIvol selected was in between 7.12 and 7.37mGy. The raw data were reconstructed using the reconstruction kernels B31f, B80f and B70f, and slice thicknesses were 1.0mm and 5.0mm. Finally, the same parameters and procedures were used for the scanning of water phantom. Friedman test and Wilcoxon-Matched-Pair test were used for statistical analysis.
RESULTS: The DLP based on the given CTDIvol values showed significantly lower exposure for protocol 4, when compared to protocol 1 (percent difference 5.18%), protocol 2 (percent diff. 4.51%), and protocol 3 (percent diff. 8.81%). The highest change in Hounsfield Units was observed with dual-energy Sn140-kV (Hounsfield unit 15.18) compared to protocol 2 (24.35HU). The differences in noise between the different clinical protocol data sets were statistically significant [protocol 3 vs. dual-energy 100-kV (p<0.01) and protocol 3 vs. dual-energy Sn140-kV (p<0.01)]. The dual-energy Sn140-kV protocol shows the highest image noise (14.5HU for 5.0mm slice (B31f) and 162HU for 1.0mm slice (B70f) thickness). The difference between reconstruction kernel B31f and B80f images made using 5.0mm reconstruction thickness was statistically significant (p<0.0312) and 1.0mm slice thickness shows the significance of p<0.0312 between B31f and B70f reconstructions. In both cases, the lowest image noise was obtained from B31f reconstructed images. Again the slice thickness significantly affects image noise (p<0.03) and the noise was higher at 1.0mm compared to that at 5.0mm slice thickness.
CONCLUSION: The clinical protocol, reconstruction kernel, slice thickness and phantom diameter or the density of material it contains directly affects the image quality. Dual energy protocol shows the lowest dose-length-product compared to all other protocols examined, the fused image shows excellent image quality and the noise is same as that of single or high-pitch mode protocol images. Advanced CT technology improves image quality and considerably reduces radiation dose.