| 胡云龙,彭婉琳,刘科伶,等.CT心肌灌注联合深度学习重建算法在冠心病临床应用中的可行性研究[J].中华放射医学与防护杂志,2026,46(5):452-457.Hu Yunlong,Peng Wanlin,Liu Keling,et al.The feasibility study on the clinical application of CTP combined with deep learning image reconstruction algorithm in coronary artery disease[J].Chin J Radiol Med Prot,2026,46(5):452-457 |
| CT心肌灌注联合深度学习重建算法在冠心病临床应用中的可行性研究 |
| The feasibility study on the clinical application of CTP combined with deep learning image reconstruction algorithm in coronary artery disease |
| 投稿时间：2025-08-26 |
| DOI：10.3760/cma.j.cn112271-20250826-00311 |
| 中文关键词: 计算机体层成像 心肌灌注 冠状动脉疾病 深度学习 |
| 英文关键词:Computed tomography Myocardial perfusion Coronary artery disease Deep Learning |
| 基金项目:国家自然科学基金(82371927);四川大学华西医院学科卓越发展1·3·5工程项目(ZYGD23024) |
|
| 摘要点击次数: 2736 |
| 全文下载次数: 37 |
| 中文摘要: |
| 目的 探讨70 kV CT心肌灌注(CTP)结合深度学习重建算法(DLIR)对图像质量及临床应用价值的影响。方法 前瞻性纳入50例疑似冠心病患者进行低剂量CTP检查,管电压70 kV,管电流200 mA,记录扫描的辐射剂量。根据冠状动脉狭窄程度,将患者分为健康对照组(n=13)、非显著狭窄组(n=18)及显著狭窄组(n=19)。所有CTP图像分别采用传统滤波反投影法(FBP)与高等级的DLIR算法进行重建。对两种算法重建的图像质量进行主客观综合评价。客观评价包括图像噪声、信噪比(SNR)以及对比噪声比(CNR);主观质量采用4分制。同时,定量分析患者整体及心肌各节段平均心肌血流量(MBF)。结果 DLIR算法重建的CTP图像噪声(37.5±3.4)较传统FBP算法重建(51.2±3.6)明显降低(t=31.41,P<0.001)。相应地,DLIR重建的图像SNR、CNR升高(t=-9.74、-8.01,P<0.001)。传统FBP算法与DLIR算法测量所得的心肌节段平均MBF差异无统计学意义。健康对照组、非显著狭窄组及显著狭窄组的平均MBF分别为(137.72 ± 4.26)、(132.67 ± 4.98)、(120.02 ± 8.47) ml·100 ml -1·min-1(F=33.61,P<0.001)。同样,狭窄程度越高的冠状动脉供血心肌测得的MBF越低(F=74.21,P<0.001)。结论 70 kV CTP结合DLIR算法可改善图像质量,DLIR算法的使用并不影响MBF计算,且能有效评估CAD患者的血流动力学状态,具有较高的临床应用价值。 |
| 英文摘要: |
| Objective To investigate the image quality and clinical value of the 70 kV myocardial computed tomography perfusion (CTP) combined with deep learning image reconstruction (DLIR) algorithm for patients with coronary artery disease (CAD). Methods Totally 50 consecutive patients with suspected CAD were enrolled for low-dose CTP using 70 kV and 200mA. The radiation dose was recorded. According to the degree of coronary artery stenosis, the patients were divided into the control group (n=13), the non-significant stenosis group (n=18) and the significant stenosis group (n=19) All CTP images were reconstructed using both filtered backprojection (FBP) and high-level DLIR algorithm. The image quality of the two groups was evaluated subjectively and objectively.The subjective quality was scored using a 4-point scale. And objective evaluations included image noise, signal-to-noise ratio (SNR), and contrast-to-noise ratio (CNR). The mean myocardial blood flow (MBF) and MBF of each myocardial segment of the patients were analyzed using one-way ANOVA. Results Compared with the CTP images reconstructed by the traditional FBP algorithm (51.2±3.6), those reconstructed by the DLIR algorithm (37.5±3.4) was significantly reduced(t=31.41, P<0.001). The SNR and CNR of the DLIR group increased correspondingly (t=-9.74, -8.01, P<0.001). There was no statistically difference in the average MBF of myocardial segments measured by the traditional FBP and DLIR algorithm. The average MBF of the control group, non-significant stenosis group and significant stenosis group were (137.72±4.26), (132.67±4.98) and (120.02±8.47) ml·100 ml -1·min-1, showed a statistically significant difference (F=33.61, P<0.001). Similarly, the MBF decreased among territories supplied by coronary arteries with higher stenosis (F=74.21, P<0.001). Conclusions 70 kV CTP combined with DLIR algorithm can improve image quality without affecting on MBF calculation. This approach effectively assesses the hemodynamic status of CAD patients and demonstrates high clinical application value. |
| HTML 查看全文 查看/发表评论 下载PDF阅读器 |
| 关闭 |
|
|
|