医学
放射科
数字减影血管造影
磁共振血管造影
计算机断层血管造影
血管造影
磁共振成像
腰骶关节
椎动脉
瘘管
核医学
外科
作者
Jinyu Zhu,Wangshu Zhu,Ming‐Hua Li,Xiaoer Wei,Zheyi Chen,Yuehua Li
标识
DOI:10.1016/j.wneu.2025.123745
摘要
OBJECTIVE: To investigate the performance of time-resolved angiography with stochastic trajectories (TWIST-MRA) and dual-source CT angiography (CTA), applied independently and in combination, for preoperative assessment of spinal vascular malformation (SVM). METHODS: This retrospective study included 77 patients with suspected SVM who underwent TWIST-MRA, CTA, and digital subtraction angiography between September 2019 and December 2022. Precision recall curve analysis was used to assess diagnostic performance. The feeding artery and fistula location were evaluated using radiologist confidence level by CTA and TWIST-MRA. RESULTS: Among the 77 patients, 71 had digital subtraction angiography-confirmed SVM (cervical, n = 7; thoracolumbar, n = 60; deep lumbosacral, n = 4) and 6 did not have SVM. Both TWIST-MRA and CTA showed excellent accuracy (96.1% vs. 94.8%) and sensitivity (98.6% vs. 97.2%) for diagnosis of SVM subtypes. TWIST-MRA performed slightly better than CTA in identifying the feeding artery and fistula (area under the curve-precision recall, 0.958 vs. 0.944); however, sensitivity was not statistically different (P = 0.512). Among SVM patients, the radiologist confidence levels in identifying feeding artery and fistula of SVMs were significantly higher with CTA+TWIST-MRA than with TWIST-MRA alone (P = 0.007). The radiologist confidence levels were significantly higher with CTA alone and CTA+TWIST-MRA compared to TWIST-MRA alone for fistula identification of cervical and deep lumbosacral SVMs (P < 0.001). CONCLUSIONS: Both TWIST-MRA and CTA are reliable for diagnosing SVM and localizing the feeding artery and fistula. While TWIST-MRA remains the primary technique, the addition of CTA can improve diagnostic confidence. CTA may be considered as an alternative to TWIST-MRA for SVMs in the cervical and lumbosacral regions.
科研通智能强力驱动
Strongly Powered by AbleSci AI