Coefficient Design and Recognition Optimization for Dual-energy Cranial CT Non-linear Fusion Technology

WU Jing-qiang; LAN You-xi; HE Yong-hong

摘要：Objective:This study aims to explore the optimal selection of blending coefficients for non-linear blending(NLB)technology in dual-source dual-energy cra-nial CT angiography(CTA)to enhance image quality and contrast.Methods:A total of 48 dual-energy head CT images with a subjective score of less than 3 points in the initial linear fusion image quality were selected for NLB processing.The fixed CTA energy was 70 and 140 kV,and the fusion center λ(0,80,150,250 HU),fusion width ω(0,100,200,300 HU)and total contrast agent(20,22,24,25 mL)were analyzed by comparing the cerebrovascular area of interest of the experimental subjects for image quality scor-ing,and the response surface method was used to determine the optimization parameters of the non-linear fusion technology of dual-energy cranial CTA.Results:According to the mean values of each factor in the orthogonal test,it was found that λ had the highest influence level in K2,ω had the highest influence level in K1,and the total amount of contrast agent K3 had the highest influence level.Range analysis revealed the order of influence as λ>contrast agent volume>ω.Through response surface analysis,the optimal parameters were identified as a blending center λ=125 HU,blending width ω=150 HU,and contrast agent volume=22.50 mL.The average subjective score of LB images was(2.62±0.49),and the subjective image quality scores were higher than those of LB images after NLB processing,with subjective quality scores of(3.79±0.65),with t=21.462 and P=0.000.Conclusion:For dual-energy cranial CTA,the optimal coefficient design for NLB technology can be achieved by selecting λ=125 HU,ω=150 HU,and a contrast agent volume of 22.50 mL,thereby improving contrast-to-noise ratio(CNR)and overall image quality.