摘要
目的 探讨人工智能后处理在"双低"头颈部CT血管成像(computed tomography angiogra-phy,CTA)中的临床价值.方法 选取2023年1-7月聊城市人民医院CT室行"双低"(低管电压、低碘摄入量)头颈部CTA检查患者的113份成像片为研究对象,对原始CTA图像进行常规重建并进行图像后处理.根据后处理方式分为人工组(CTA薄层图像传送至GE AW4.7后处理工作站进行人工手动图像后处理)与人工智能(artificial intelligence,AI)组(CTA薄层图像传至头颈AI智能辅助诊断系统上进行图像优化及血管分割,生成"双低"CTA图像进行AI后处理).对比两组图像质量(客观评价、主观评价)、图像处理时间.结果 AI组图像质量主观质量评分中图像对比度质量评分为(3.56±0.51)分、噪声质量评分为(3.05±0.51)分、伪影质量评分为(3.21±0.15)分、几何畸变评分为(3.22±0.55)分,均高于人工组的(3.01±0.38)分、(2.87±0.31)分、(3.12±0.15)分、(2.99±0.58)分,差异有统计学意义(t=2.912、3.206、4.510、3.059,P均＜0.05).AI组处理图像处理时间[(5.76±1.63)min]短于人工组[(12.03±1.47)min],AI组图像处理效率明显高于人工组,差异有统计学意义(t=30.366,P＜0.05).结论 人工智能后处理在"双低"头颈部CT血管成像中应用,可提高图像质量、减少噪声,并显著缩短图像处理时间,为临床提供更有效的辅助诊断手段.
Abstract
Objective To explore the clinical value of artificial intelligence post-processing in"double-low"head and neck computed tomography angiography(CTA).Methods 113 imaging slices of patients who un-derwent"double-low"(low tube voltage,low iodine intake)head and neck CTA in the CT room of Liaocheng People's Hospital from January to July 2023 were selected as the study subjects.The original CTA images were routinely reconstructed and post-processed,and according to the post-processing method,they were cat-egorized into the manual group(CTA thin-layer images were transferred to a GE AW4.7 post-processing work-station for manual image post-processing)and the artificial intelligence(AI)group(CTA thin-layer images were transferred to the Head and Neck AI Intelligent Assisted Diagnostic System for image optimization and vascular segmentation,and the"double-low"CTA images were generated for AI post-processing).The image quality of the two groups was compared(objective evaluation,subjective evaluation),and the image processing time.Results The subjective quality scores of image quality in the AI group were higher than those of the manual group in image contrast quality score was(3.56±0.51)points,noise quality score was(3.05±0.51)points,artifact quality score was(3.21±0.15)points,and geometrical aberration score was(3.22±0.55)points,which were higher than those of the manual group(3.01±0.38)points,(2.87±0.31)points,(3.12±0.15)points,(2.99±0.58)points,and the difference was statistically significant(t=2.912,3.206,4.510,3.059,all P＜0.05).The pro-cessing time of the AI group was shorter[(5.76±1.63)min]than that of the manual group[(12.03±1.47)min],and the image processing efficiency of the AI group was significantly higher than that of the manual group,and the difference was statistically significant(t=30.366,P＜0.05).Conclusion Artificial intelligence post-processing has the application in"double-low"head and neck CT angiography,which can improve image qual-ity,reduce noise,and significantly shorten the image processing time,providing a more effective auxiliary diag-nostic means for the clinic.
万方数据