摘要
目的探讨深度学习图像重建(DLIR)算法联合低管电压(70 kVp)在头颈部CT血管成像中的应用价值。资料与方法回顾性收集2021年11月—2022年11月宁夏回族自治区人民医院行头颈部CT血管成像检查的40例患者,管电压70 kVp,管电流智能调控SmartmA模式。对原始数据进行自适应统计迭代重建(ASiR-V60%、ASiR-V90%)及DLIR(DLIR-L、DLIR-M、DLIR-H),比较不同算法及水平重建图像的主观及客观评价指标。结果颈部及颅内动脉不同水平ASiR-V及DLIR重建图像CT值差异均无统计学意义(P>0.05)。随着ASIR-V及DLIR重建水平增加,颈部及颅内图像噪声均减低,DLIR-H与ASiRV60%差异均有统计学意义(P<0.05),分别减低24.30%、29.42%;颈部及颅内图像信噪比(SNR)、对比噪声比(CNR)均升高,DLIR-H重建图像SNR、CNR最高。颈总动脉分叉部、颈内动脉C4段、椎动脉V4段SNR、CNR与ASiR-V60%差异均有统计学意义(P<0.05),SNR分别提高55.60%、43.90%、44.66%,CNR分别提高55.57%、44.24%、45.10%;大脑中动脉SNR、CNR较ASiR-V60%提高45.39%、45.89%,差异有统计学意义(P<0.05)。ASiR-V90%主观评分较ASiR-V60%降低,差异无统计学意义(P>0.05)。DLIR主观评分随重建水平升高而升高,DLIR-H较DLIR-M、DLIR-L差异有统计学意义(P均<0.05),DLIR-H及DLIR-M的主观评分高于ASiR-V60%与ASiR-V90%,差异有统计学意义(P均<0.05)。结论在低管电压头颈部CT血管成像中,与ASiR-V相比,DLIR可以进一步降低图像噪声,提升图像质量和诊断信心,其中DLIR-H表现最佳。
Purpose To explore the application value of deep learning image reconstruction(DLIR)algorithm combined with low tube voltage(70 kVp)in head and neck CT angiography imaging.Materials and Methods Retrospective analysis was performed on 40 patients who underwent head and neck CT angiography examination in People's Hospital of Ningxia Hui Autonomous Region from November 2021 to November 2022,the scanning tube voltage was 70 kVp and the current was in SmartmA mode.The original data were reconstructed with different algorithms and levels,including adaptive statistical iterative reconstruction-veo(ASiR-V60%and ASiR-V90%)and DLIR(DLIR-L,DLIR-M,DLIR-H).The subjective and objective evaluation of different algorithms and levels reconstructed images were compared.Results CT values of ASiR-V and DLIR reconstruction images at different levels in neck and intracalvarium had no statistical significance(P>0.05).With the increase of ASiR-V and DLIR reconstruction level,the image noise of neck and intracalvarium was reduced,compared with ASiR-V60%,DLIR-H decreased 24.30%and 29.42%,respectively(P<0.05).With the increase of ASiR-V and DLIR levels,signal to noise ratio(SNR)and contrast-to-noise ratio(CNR)of neck and intracalvarium images increased,the SNR and CNR of DLIR-H were the highest.The SNR and CNR of common carotid artery bifurcations,C4 segment of internal carotid artery and V4 segment of vertebral artery were statistically significant compared with ASiR-V60%(P<0.05),SNR increased by 55.60%,43.90%,44.66%,CNR increased by 55.57%,44.24%,45.10%,respectively.The SNR and CNR of middle cerebral artery were 45.39%and 45.89%higher than that of 6 ASiR-V60%,with statistical significance(P<0.05).The subjective score of ASiR-V90%was lower than that of ASiR-V60%,and there was no statistical significance(P>0.05).The subjective score of DLIR increased with the level of reconstruction,and DLIR-H was significantly higher than DLIR-M and DLIR-L(P<0.05),the subjective score of DLIR-H and DLIR-M was significantly higher than ASiR-V60%and ASiR-V90%(P<0.05).Conclusion In low tube voltage head and neck CT angiography imaging,compared with ASiR-V,DLIR can further reduce image noise,improve image quality and diagnostic confidence,among which DLIR-H performs best.
作者
杨彦兵
阮小伟
王泽润
于梓婷
杨利莉
汪芳
YANG Yanbing;RUAN Xiaowei;WANG Zerun;YU Ziting;YANG Lili;WANG Fang(The Medical Imaging Center,People's Hospital of Ningxia Hui Autonomous Region,Ningxia Medical University,Yinchuan 750001,China)
出处
《中国医学影像学杂志》
CSCD
北大核心
2024年第6期553-558,共6页
Chinese Journal of Medical Imaging
基金
2022年宁夏回族自治区自然科学基金(2022AAC03390)
2022年宁夏回族自治区卫生健康系统科研课题(2022-NWKY-015)。
关键词
头颈部CT血管成像
图像质量
深度学习图像重建
迭代重建
Head and neck CT angiography
Image quality
Deep learning image reconstruction
Iterative reconstruction
