Tanahashi Yukichi, Kondo Hiroshi, Yamamoto Masayoshi, Osawa Marie, Yokoyama Taro, Sugawara Toshimasa, Kawada Hiroshi, Goshima Satoshi, Matsuo Masayuki, Furui Shigeru, Oba Hiroshi
Department of Radiology, Gifu University School of Medicine, 1-1 Yanagido, Gifu, 501-1194, Japan.
Department of Radiology, Teikyo University School of Medicine, 2-11-1 Kaga, Itabashi, Tokyo, 173-8605, Japan.
Cardiovasc Intervent Radiol. 2018 Nov;41(11):1786-1793. doi: 10.1007/s00270-018-2027-8. Epub 2018 Jul 10.
To assess the reliability of a prototype automated supplying artery tracking software (ASATS) using multidetector-row CT (MDCT) images in emergent TAE.
Consecutive 53 patients underwent 57 sessions of emergent TAE during 7 months. Twenty-one cases were excluded due to a lack of CT data (n = 12) or negative angiographic findings (n = 9). Remaining 34 sessions of TAE and MDCT images in 32 patients (mean age 62.9 years; age range 37-92 years) were enrolled. ASATS was retrospectively conducted for the identification of supplying arteries which were confirmed with angiography (automated method). Manual modification was added as needed (semi-automated method). Two observers independently reviewed the MDCT images to detect supplying arteries (manual method). Detectability of supplying artery and time to analysis were compared among the automated, semi-automated, and manual methods by both observers.
A total of 64 bleeding sites were demonstrated on angiography. The detectability was 28 (43.8%) for automated method, 53 (82.8%) for semi-automated method, 55 (85.9%) for observer 1, and 58 (90.6%) for observer 2. Detectability of semi-automated method was significantly better than of automated method (P = 0.000) and comparable with manual method by both observers (P = 0.193 and 0.081). Average time to analysis was 185.4 s for automated method, 297.2 s for semi-automated method, 186.2 s for observer 1, and 243.7 s for observer 2.
ASATS has a sufficient ability to identify supplying arteries of bleeding by adding manual modification as needed and can be used for emergent TAE.
Level 4, Case Control Study.
评估一款使用多排探测器CT（MDCT）图像的原型自动供血动脉追踪软件（ASATS）在急诊经动脉栓塞术（TAE）中的可靠性。
7个月内连续53例患者接受了57次急诊TAE。21例因缺乏CT数据（n = 12）或血管造影阴性结果（n = 9）被排除。纳入了32例患者（平均年龄62.9岁；年龄范围37 - 92岁）的34次TAE及MDCT图像。对ASATS进行回顾性分析以识别经血管造影证实的供血动脉（自动方法）。必要时进行手动修正（半自动方法）。两名观察者独立查看MDCT图像以检测供血动脉（手动方法）。两名观察者对自动、半自动和手动方法的供血动脉可检测性及分析时间进行比较。
血管造影共显示64个出血部位。自动方法的可检测性为28个（43.8%），半自动方法为53个（82.8%），观察者1为55个（85.9%），观察者2为58个（90.6%）。半自动方法的可检测性显著优于自动方法（P = 0.000），且与两名观察者的手动方法相当（P = 0.193和0.081）。自动方法的平均分析时间为185.4秒，半自动方法为297.2秒，观察者1为186.2秒，观察者2为243.7秒。
ASATS通过按需添加手动修正，有足够能力识别出血的供血动脉，可用于急诊TAE。
4级，病例对照研究。
