Research on correlation of CT iterative reconstruction algorithm with CT image quality and radiation dosage

Objective: To explore the influences of multi-slice spiral CT iterative reconstruction algorithm on image quality and exposure dose. Methods: Using both algorithms-iterative reconstruction algorithm (IR) and filtered back-projection reconstruction algorithm (FBP) on the Philips Brilliance 64 CT scanner iDose V3.5 platform, the homogeneity, level of noise and resolution ratio of the Catphan 500 model were acquired and compared between the two groups using Wilcoxon test. Routine CT images of 20 abdomen, 20 skull and 20 temporal bone were achieved and reconstructed using IR as IR group. Sixty routine CT images were randomly selected and divided into 3 groups with 20 in each group, which were reconstructed using FBP as FBP group. Wilcoxon test was used to compare the image quality grades and the radiation does between these two groups. Results: Model scanning: With CTDIvol of 5.6, 8.0, 16.1, 24.1 and 32.2 mGy, the noise of IR group (1.45%, 0.96%, 0.68%, 0.53% and 0.47% Respectively) were lower than those of FBP group (1.74%, 1.14%, 0.77%, 0.61% and 0.53% Respectively) (Z = -2.023, P = 0.043). There were no statistical differences in homogeneity, high-contrast resolution and low-contrast resolution between the two groups (P > 0.05). (2) Clinical patients: the abdominal CTDIvol of FBP and HR groups were (16.89 ± 2.19) mGy and (11.06 ± 4.15) mGy, respectively. CTDIvol of IR group was 34.52% lower with statistic difference (Z = -3.398, P =0.001). The objective scores were (3.75 ± 0.34) and (3.59 ± 0.39), respectively. There was no statistical difference (Z = -1.660, P = 0.097), Images of both IR and FBP group could clearly show brain structures and lesions, CTDIvol of FBP group was (48.23 ± 2.59) mGy; CTDIvol of IR group was (43.82 ± 1.09) mGy using 1^st level iterative reconstruction and (42.91 ± 0.81) mGy using 5^th level iterative reconstruction. There were statistical differences between FBP group and IR group using 1^st level and 5^th level (Z = - 3.775 and -3.812, P = 0.001). The image quality grade of FBP group was (3.63 ± 0.45), and that of IR group using 1^st level iterative reconstruction was (3.68 ± 0.40). There was no statistical difference (Z = -1.660, P = 0.097). The image quality grade of IR group using 5^th level iterative reconstruction was (1.83 ± 0.33, there was statistical difference (Z = -3.952, P = 0.001). The temporal CTDIvol of FBP group and IR group were (57.97 ± 1.89) mGy and (28.31 ± 0.63) mGy, respectively, CTDIvol of IR group was 51.16% lower with statistic difference (Z = -4.300, P = 0.001). The contrast and acutance of images of both groups decreased. The subjective score were (4.40 ± 0.73) and (3.90 ± 0.76 ), respectively; there was statistical difference (Z = -2.055, P = 0.040 ). Conclusions: Under the condition that the images quality meets the demand of diagnosis, CT iterative reconstruction algorithm may effectively reduce exposure dose. The iterative level should be selected in step with anatomic characteristics. (authors)