Impact Of Deep Learning Image Reconstruction Algorithms On CT Radiomic Features In Patients With Liver Malignant Tumors

Objective:To evaluate the impact of deep learning image reconstruction(DLIR)algorithm and adaptive statistical iterative reconstruction-Veo(ASIR-V)algorithm compared with filtered back projection(FBP)reconstruction algorithm on abdominal CT radiomic features acquired in portal venous phase in patients with liver malignant tumor,and explore the difference of the influence degree of different reconstruction strengths of different reconstruction algorithms on CT radiomic features.Materials and Methods:Sixty patients with liver malignant tumors who underwent contrast-enhanced abdominal CT were retrospectively enrolled from July 2020 to November 2020.Twentyone patients with primary liver malignant tumors,including seventeen cases of hepatocellular carcinoma and four cases of cholangiocarcinoma,and thirty-nine patients with secondary liver metastasis.There were 41 males and 19 females,ranging in age from 32 to 90 years with an average age of(64.92 ± 12.80)years.Six groups including filtered back projection(FBP),ASIR-V(30%,70%)and DLIR at low(DLIR-L),medium(DLIR-M)and high(DLIR-H),were reconstructed from original scan datasets in portal venous phase by using GE Medical DLIR reconstruction engine.The 3D slicer 4.11 software was used to segmented the region of interests(ROI)of 2D and 3D liver tumors,peritumor and liver parenchyma,and the Pyradiomics package in 3D Slicer software was used to extract the radiomic features.The extraction included18 first-order features,75 texture features and 144 wavelet features,totaling 5688 features.Two radiologists delineated the 3D lesions separately,and one of them repeated the delineation in a month.SPSS 25.0 software was used to analyze the extracted CT radiomic features.In accordance with the Kolmogorov-Smirnov and Levene’s test results,One-way analysis of variance(ANOVA)or Kruskal-Wallis test was applied for comparing features among the six groups.Least Significant Difference(LSD)or Bonferroni method was used for further comparison and P value was adjusted using the Bonferroni correction.P < 0.05 indicated that the difference was statistically significant.The consistency of 3D lesion feature extraction was assessed by calculating intraclass correlation coefficient(ICC).ICC < 0.40 was considered as poor consistency,0.40 ≤ ICC ≤ 0.75 was considered as moderate consistency,and ICC > 0.75 was considered as excellent consistency.Results:1.Different reconstruction algorithms influenced most radiomic features.The percentages of first-order,texture and wavelet features without statistical difference among 2D and 3D lesions,peritumor and liver parenchyma for all six groups were 27.78%(5/18),5.33%(4/75)and 5.56%(1/18),respectively(all p > 0.05).2.With the increase of ASIR-V and DLIR reconstruction strength,the number of features without statistical difference decreased.Compared with FBP,the radiomic features of ASIR-V30% and 70% without statistical difference decreased from 71.31% to23.95%,and DLIR-L,DLIR-M,and DLIR-H decreased from 31.65% to 27.11% and23.73%.3.Among texture features,unaffected features of peritumor(20/75)were larger than those of lesions and liver parenchyma,and unaffected 3D lesions features(11/75)were larger than those of 2D lesions(7/75).Only four features were not significantly different between groups in liver lesions,peritumor and liver parenchyma(all p > 0.05).4.In wavelet features,the number of unaffected features through wavelet-LLL filtering transform(40/72)was larger than other wavelet filtering methods.5.The consistency of 3D lesion first-order features was excellent,with intra-and inter-observer ICCs ranging from 0.891 to 0.999 and 0.880 to 0.998.Conclusions:1.Both ASIR-V and DLIR algorithms with different strengths influenced the radiomic features of abdominal CT images in portal venous phase.2.The influence of different reconstruction algorithms on the radiomic features aggravated as reconstruction strength increased.3.For the texture features,the number of unaffected features of peritumor were more than those of liver lesions and liver parenchyma,and the number of unaffected features of 3D lesions were larger than those of 2D lesions.4.In liver lesions,peritumor and liver parenchyma,mean was not significantly different between groups both before and after wavelet filtering transformation.5.The number of unaffected features through wavelet-LLL filtering transform was larger than other wavelet filtering methods.