目的：应用不同的卷积核算法，使用高级血管分析软件的半自动追踪曲面重组功能显示颈内动脉颅内段及椎动脉，探讨最容易曲面重组与骨质结构关系密切血管的卷积核算法。方法对32例临床疑有脑缺血或蛛网膜下腔出血病例进行头颈部Neuro DSA CTA扫描。默认图像重建卷积核算法为H10f，然后分别选择H48f、H50f及H60f卷积核算法进行重建。使用Inspace软件下的AVA软件进行血管分析，计数载入不同卷积核算法图像半自动追踪成功曲面重组颈内动脉及椎动脉支数。结果32例共128支血管。H10f算法图像半自动追踪曲面重组显示0支血管，成功率为0；H48f算法图像显示44支血管，成功率为34%；H50f算法图像显示68支血管，成功率为53%；H60f算法图像显示108支血管，成功率为84%，两两比较差异均有统计学意义(P<0.01)；H60f算法图像半自动追踪曲面重组成功率明显高于其他两种，H60f算法图像对颈内动脉曲面重组成功率明显高于椎动脉。结论使用AVA软件对颈内动脉颅内段及椎动脉进行血管分析时，应用较高卷积核(H60f)算法能明显改善半自动追踪血管曲面重组成功率，利于进一步进行血管分析，有较大临床实用价值。%Objective To investigate the correct kernel algorithm for vessels close to adjacent bone when semi-automated tracing internal carotid artery (ICA) intracranial segment and vertebral artery (VA) with curved planar reformation (CPR) in advanced vessels analysis (AVA). Methods Thirty-two patients were enrolled and were scanned by Neuro DSA CT angiograply. Algorithms of H10f, H48f, H50f, H60f were selected. Inspace AVA was applied for ves-sels analysis, the number of ICA and VA was counted, which were semi-automated traced well. Results The 32 pa-tients had a total of 128 vessels. No vessel was successfully curve planar reformatted with H10f. Forty-four vessels (34%) was successfully displayed with H48f and 68 vessels (53%) with H50f, 108 vessels (84%) with H60f. Signifi-cant difference was found between different groups (P<0.05). The achievement ratio of H60f was significantly higher than H48f and H50f, and achievement ratio of H60f was higher in ICA than VA. Conclusion High kernel algorithm could improve achievement ratio in semi-automated traced CPR for advanced vessel analysis, when analyzing ICA in-tracranial segment and VA with AVA.
展开▼
