目的 探讨CT能谱单能量成像联合图像融合技术对小肝细胞癌图像质量的影响.方法 回顾性分析经手术病理证实为肝细胞癌,行CT能谱成像双期增强扫描,且肿瘤病灶最大径≤3 cm的30例患者(40个小肝细胞癌病灶).患者均行上腹部CT平扫和双期能谱CT增强扫描.应用能谱分析软件,重建出140 kVp常规混合能量图像(A组)和单能量图像(40~ 140 keV),并利用图像融合软件将对比噪声比(CNR)最高时的最佳单能量图像(B组)与图像噪声最低时的70 keV图像(C组)进行融合后获得融合图像(D组).对4组图像质量进行客观评价(测量计算40个病灶CNR、噪声)和主观评分(包括30例患者图像噪声评分、总体图像质量评分和病灶显著性评分).采用单因素方差分析和Kruskal-WallisH检验比较4组图像的客观和主观评价指标.结果 A、B、C、D组病灶的CNR分别为1.3±0.6、2.1±0.6、1.5±0.6和2.4±1.3,病灶噪声分别为(20±7)、(32±9)、(18±3)、(24±6)HU,差异均有统计学意义(F值分别为5.724、13.619,P均<0.01).CNR值D组高于A组和C组(P均<0.05),和B组的差异无统计学意义(P>0.05);图像噪声值D组低于B组,高于C组(P均<0.05),和A组的差异无统计学意义(P>0.05).4组图像的噪声评分分别为(3.5±0.5)、(3.3±0.4)、(3.6±0.5)、(3.5±0.4)分,总体图像质量评分分别为(3.2±0.4)、(3.3±0.3)、(3.1±0.3)、(3.7±0.4)分,病灶显著性评分分别为(3.3±0.4)、(3.9±0.4)、(3.2±0.4)、(3.7±0.4)分,差异均有统计学意义(Z值分别为9.581、37.495和43.436,P均<0.05).4组图像评分均>3分,符合临床诊断的要求.D组的图像噪声评分高于B组,总体图像质量评分D组高于其他3组,病灶显著性评分D组高于A组(P均< 0.05).结论 CT能谱单能量成像联合图像融合技术可以在提高总体图像质量的同时保持或提高小?
Objective To assess the effects of image fusion of CT spectral monochromatic imaging on image quality in small hepatocellular carcinoma (HCC).Methods Thirty patients with 40 pathologically proven small HCCs (≤3 cm) underwent upper abdominal plain CT and dual-phase enhanced spectral CT scan were analyzed retrospectively.Conventional 140 kVp polychromatic images (group A) and monochromatic images with energy levels from 40 to 140 keV were reconstructed by using spectral imaging viewer.Monochromatic images with highest CNR (group B)and 70 keV images with lowest noise (group C) were fused to generate fused images (group D) with image fusion software.Objective evaluation of 40 HCCs [contrast-to-noise ratio (CNR) of lesion,image noise in HU] and subjective rating score of 30 patients (image noise score,overall image quality score,and lesion conspicuity score) among the four groups were compared by using One-way ANOVA and Kruskal-Wallis H test.Results CNR (1.3±0.6,2.1±0.6,1.5±0.6 and 2.4± 1.3 respectively) and image noise [(20±7),(32±9),(18±3) and (24±6) HU respectively] among group A,B,C and D all had statistical differences (F =5.724 and 13.619,both P values < 0.01).CNR in group D was higher as compared with group A and C (both P values < 0.05),but was similar to group B (P > 0.05).Image noise in group D was lower than group B and higher than group C,but showed no difference from group A (all P values > 0.05).Image noise scores [(3.5±0.5),(3.3±0.4),(3.6±0.5)and(3.5±0.4)point,respectively],overall image quality scores [(3.2 ± 0.4),(3.3 ± 0.3),(3.1 ± 0.3) and (3.7 ± 0.4) point respectively] and lesion conspicuity scores [(3.3 ±0.4),(3.9±0.4),(3.2±0.4) and (3.7 ±0.4) point,respectively] among the four groups all had statistical differences (Z =9.581,37.495 and 43.436,all P values < 0.05).Scores of the four groups were all greater than 3 and met the clinical diagnostic lev