Research On The Combination Of Model-based Knowledgebased Iterative Reconstruction（IMR） And Orthopedic Metal Artifact Reduction（O-MAR） Used In The Computed Tomography Image Evaluation Of The ¹²⁵I Seeds Implantation Therapy

Objective: Explore the application value of the combination of Model-Based KnowledgeBased Iterative Reconstruction(IMR)and orthopedic metal artifact reduction(OMAR)technique in improving image quality and reducing the artifact caused by 125 I.Methods: 1.Patients: twenty-four cases have been collected,of which the patients received the implantation of 125 I of retroperitoneal from Dec.2015 to Jun.2017,and got post-operation follow-ups of 256 CT.There are 15 females and 9 males,who aged 42 to 79,and the average age is 55.8.2.CT Scanning and image postprocessing: 2.1 Scanning Equipment and Parameters: The scanning equipment is MSCT(Philips Brilliance i CT).Parameters: tube voltage 120 k V,Collimation Width 128× 0.625 mm,pitch 0.91 mm,matrix 512×512.2.2 Image Reconstruction: All the raw data are processed and analyzed by Philips postprocessing workstation(EBW4.01).Image reconstruction using the raw data is conducted by IMR(level 2),i Dose4(level 4)of hybrid iterative reconstruction(HIR)and FBP respectively.Each reconstruction is conducted both in with O-MAR style and without O-MAR style.Relevant data are acquired by using multi-planar reconstruction(MPR)to further process the thin slice image.3.Evaluation of image quality:3.1 Evaluation the objective quality of image: Observe and measure the ROIs on the image slice having maximal artifacts of the six reconstruction methods respectively,the mean CT value and noise value of artifacts and background noise value were measured.Calculate the artifact index(AI)and Contrast-to-noise Ratio(CNR)according to the relevant formula.Select the display slice of the maximal radius of 125 I particle and measure the length.3.2 Evaluation the subjective quality of image: The reconstructed images are scored by two experienced radiologists in respect of the artifacts of 125 I particle,the sharpness of 125 I particle and the displaying condition of tumor and the surrounding normal tissue.Perform a consistency check about the results.4.Statistical Analysis: Statistical analysis is conducted by SPSS 24.0.All the continuous variables are tested on normality and homogeneity of variance.The normal distribution data is represented by mean value ± standard deviation(x ±s),and the non-normal distribution data is represented by median and interquartile range.The consistency check of the scores given by two radiologists is performed via Kappa Check.Subjective scores are checked by using the Friedman test,and if significant differences are found,then use Wilcoxon Rank Test to perform pairwise comparison.Objective scores are checked by using means of repeated measures analysis of variance,and if significant differences are found,then use paired t-test corrected by Bonferroni correction to perform pairwise comparison.Adopt the analysis of paired t-test between the particle length measured by the six methods and the real particle length respectively.If p<0.05,then the differences are statistically significant.Results: 1.Objective analysis of image quality: The image noise of IMR is higher than i Dose4 and FBP,and the difference is statistically significant(p < 0.05);When combining with O-MAR technique,The image noise of IMR,i Dose4 and FBP using O-MAR sees a significant reduction comparing with the ones without combining O-MAR,and the differences are statistically significant(p < 0.001).The artifact index of IMR has no statistically significant difference with that of i Dose4,but with that of FBP(p < 0.05).The artifact index of IMR,i Dose4 and FBP using O-MAR sees a significant reduction comparing with the ones without combining O-MAR,and the differences are statistically significant(p < 0.001).The CNR of the image constructed by IMR is much higher than that of the image constructed by IMR,and the difference is statistically significant(p < 0.05).The particle length measured in the images constructed by IMR and IMR+O-MAR has no statistically significant difference with the real particle length,can represent the real one;The particle length measured in the images constructed by i Dose4,i Dose4+OMAR,FBP and FBP+OMAR has statistically significant difference with the real particle length,cannot represent the real one.2.Subjective analysis of image quality: The Kappa value of the consistency test concerning the scores given by two radiologists is 0.60;The objective scores of the image constructed by IMR is higher than that of FBP,the difference is statically significant(p <0.05).The difference between the objective scores of the methods combining with O-MAR and that without O-MAR is not statically significant.Conclusions: 1.Comparing with i Dose4 and FBP,IMR can significantly reduce image noise,improve the objective and subjective image quality,enhancing the displaying situation of 125 I particle and the surrounding issue.2.O-MAR can significantly reduce the artifacts of 125 I particle in CT image,improve the objective image quality.3.Combining IMR with O-MAR can minimize the artifacts of 125 I particle in CT image,decrease image noise and improve the objective and subjective image quality,comparing with the other five methods.This combination can enhance the displaying situation of 125 I particle and the surrounding issue,provide a better option for the postoperation follow-up and effect evaluation after 125 I particle implantation.