• 主题
高级搜索  >
历史搜索 清空历史
首页> 外文期刊>Acta Radiologica >Computer aided diagnosis for ASPECT rating: initial experiences with the Frontier ASPECT Score software
【24h】

Computer aided diagnosis for ASPECT rating: initial experiences with the Frontier ASPECT Score software

机译：计算机辅助诊断方面评分：与前沿方面的初始经历

获取原文
获取原文并翻译 | 示例
           
页面导航
  • 摘要
  • 著录项
  • 相似文献
  • 相关主题

摘要

Background Computer-aided diagnosis is increasingly used in radiology and may support not only unexperienced readers but also senior radiologists. It appears promising, especially in the sometimes challenging detection of early ischemic changes in stroke CT. Purpose To compare the new post-processing software prototype Frontier_ASPECTS against two senior radiologists in ASPECTS evaluation. Material and Methods Retrospectively, pre-interventional CTs of 100 patients, who underwent endovascular revascularization for acute middle cerebral artery ischemia, were blindly re-analyzed with respect to ASPECTS by two neuroradiologists (separately and in consensus) and by use of Frontier_ASPECTS. In addition to a fully automatic Frontier_ASPECTS reading (Frontier_1), Frontier_ASPECTS readings subsequently manually corrected for old cerebral defects (Frontier_2a), the affected hemisphere (known from CT angiography, Frontier_2b), and both (Frontier_3) were assessed. Statistical analysis was performed by intraclass correlation and Bland-Altman analysis. Results Median ASPECTS was 10 for Frontier_3 (range = 5-10), 10 for radiologist_1 (range = 4-10), 9 for radiologist_2 (range = 2-10), and 10 for consensus reading (range = 2-10). All Frontier_ASPECTS variants correlated lowly with consensus reading (Frontier_1, r = 0.281; Frontier_2a, r = 0.357; Frontier_2b, r = 0.333; Frontier_3, r = 0.350; always P < 0.01), while both radiologists and consensus reading correlated highly (radiologist_1, r = 0.817; radiologist_2, r = 0.951; always P < 0.001). Bland-Altman analysis confirmed a worse agreement between Frontier_3 and consensus reading than between both radiologists and consensus reading. Conclusion We found only low agreement between the post-processing software Frontier_ASPECTS and expert consensus reading in ASPECTS evaluation. Notably, performance of Frontier_ASPECTS improved by simple manual corrections but is-at Frontier_ASPECTS' current development status-inferior to the performance of senior radiologists.
机译：背景技术计算机辅助诊断越来越多地用于放射学，可能不仅可以支持未经验的读者，而且可以支持高级放射科医师。它似乎有希望，特别是在有时挑战中风CT的早期缺血变化的发现。目的在方面评估中，将新的处理软件原型Frontier_Aspect进行比较两位高级放射科医师。材料和方法回顾性地，100名患有血管内血管内血管缺血的100名患者的预介入性CTS盲目地重新分析了两个神经产物（单独和共识）的方面，并通过使用Frontier_Aspects。除了完全自动的Frontier_Aspects读取（Frontier_1）外，Frontier_Aspect还针对旧的脑缺损（Frontier_2a）手动校正的读数，受影响的半球（从CT血管造影，Frontier_2b）和两者（Frontier_3）进行评估。通过脑内相关和平坦-Altman分析进行统计分析。结果中位方面为Frontier_3（范围= 5-10），10，放射科（范围= 4-10），放射科（范围= 2-10），10个用于读数（范围= 2-10）。所有Frontier_Aspects变体与共识读数（Frontier_1，R = 0.281; Frontier_2a，R = 0.357; Frontier_2b，R = 0.333; Frontier_3，R = 0.350;始终p <0.01），而辐射学家和共识读数高度（放射科_ r = 0.817;放射科医生_2，r = 0.951;始终p <0.001）。 Bland-Altman分析确认了Frontier_3与共识读数比在辐射科医生和共识读数之间进行了较差的一致性。结论我们发现后处理软件前沿和专家共识读数在方面的评估中只发现了低协议。值得注意的是，通过简单的手动校正来改善Frontier_Aspect的性能，但是Frontier_Aspects的当前开发状态​​ - 低于高级放射科医师的性能。

著录项

相似文献

  • 外文文献
  • 中文文献
  • 专利
获取原文

客服邮箱：kefu@zhangqiaokeyan.com

京公网安备：11010802029741号 ICP备案号：京ICP备15016152号-6 六维联合信息科技 (北京) 有限公司©版权所有

  • 客服微信

  • 服务号