Abstract: Objective To compare the effect of noise-optimized virtual monoenergetic imaging （VMI+） and conventional virtual monoenergetic imaging （VMI） on CT images in oral cancer patients with oral metal implants. Methods Data were retrospectively collected in 54 patients with metal implants who underwent maxillofacial scanning by using the third-generation dual-source CT in our hospital from August 2019 to April 2020. CT images with VMI+ and VMI from 40 to 190 keV （10 keV as interval） were reconstructed separately using the venous phase. CT images were analyzed byfourradiologists. The quantitative assessments regarding of CT values, noise （SD）, artifact index （AI）, and subjective assessments in hyperdense artifacts, hypodense artifacts, soft palate, floor of mouth, tongue, buccal soft tissue, subcutaneous fat, tumor, internal carotid artery （ICA）, and internal jugular vein （IJV） were compared between the VMI+ and VMI image. Results The lowest AI values were found in 120 keV VMI+ for the floor of the mouth, tongue and IJV; in 130 keV VMI+ for low-density artifacts and tumors; in 150 keV VMI+ for ICA; in 160 keV VMI+ for the buccal region; and in 190 keV VMI+ for high-density artifacts and soft palate. AI values in the VMI+ series were lower than that in all VMI series except for high-density artifacts, buccal and tumor （all P<0.05）. Image scores were highest at 190 keV VMI+ for high-density artifacts, at 180 keV VMI+ for the soft palate, at 160 keV VMI+ for the buccal region, and at 130 keV VMI+ for low-density artifacts, floor of the mouth, tongue, tumor, ICA, and IJV （all P<0.05）. The overall image quality was best at 130 keV VMI+ （all P<0.05）. Conclusion The high keV VMI+ technique of dual-source CT can effectively reduce metallic artifacts and improve image quality, and 130 keV VMI+ was the optimal energy level which could combine the best artifact removal effect and good tissue contrast.

Key words: Dual-energy, Computed tomography, Artifact, Oral cancer, Noise-optimized virtual monoenergetic imaging

摘要： 目的 对比噪声优化的虚拟单能成像（VMI+）与传统的虚拟单能成像（VMI）对具有金属植入物的口腔癌患者CT图像的影响。方法 回顾性收集2019年8月至2020年4月我院第三代双源CT行颌面部扫描且含有金属植入物的54例患者资料。使用静脉期分别重建VMI+和VMI在40 ~ 190 keV区间（10个keV为间隔）的图像。四名医生进行了定性和定量分析,包括CT值、噪声（SD）、伪影指数（AI）和主观评估,并对VMI+和VMI图像中的高密度伪影、低密度伪影、软腭、口底、舌、颊部软组织、皮下脂肪、肿瘤、颈内动脉（ICA）和颈内静脉（IJV）进行了比较。结果 在120 keV VMI+中,口底、舌和IJV的AI值最低;在130 keV VMI+中,低密度伪影和肿瘤的AI值最低;在150 keV VMI+中,ICA的AI值最低;在160 keV VMI+中,颊部的AI值最低;在190 keV VMI+中,高密度伪影和软腭的AI值最低。与所有VMI系列相比,VMI+系列的AI值较低,除了高致密伪影、颊部和肿瘤的AI值（均P<0.05）。高密度伪影的图像评分在190 keV VMI+最高,软腭在180 keV VMI+最高,颊部在160 keV VMI+最高,低密度伪影、口底、舌、肿瘤、ICA和IJV在130 keV VMI+最高（均P<0.05）。在130 keV VMI+时,整体图像质量是最好的（均P<0.05）。结论 双源CT的高keV VMI+技术能有效减少金属伪影并提高图像质量,130 keV VMI+图像能兼顾最佳伪影去除效果及良好的组织对比。

关键词: 双能量, 计算机断层扫描, 伪影, 口腔癌, 噪声优化的虚拟单能成像技术