| 许晨阳,许峥,王梅云,李志洁,文泽军.基于不同重建参数光子计数CT预测胃癌T分期[J].中国介入影像与治疗学,2026,23(6):376-380 |
| 基于不同重建参数光子计数CT预测胃癌T分期 |
| Photon-counting CT based on different reconstruction parameters for predicting T-stage of gastric cancer |
| 投稿时间：2026-03-23 修订日期：2026-05-11 |
| DOI：10.13929/j.issn.1672-8475.2026.06.011 |
| 中文关键词: 胃肿瘤 光子 体层摄影术，X线计算机 卷积核 量子迭代重建 |
| 英文关键词:stomach neoplasms photons tomography,X-ray computed convolution kernel quantum iterative reconstruction |
| 基金项目: |
|
| 摘要点击次数: 0 |
| 全文下载次数: 0 |
| 中文摘要: |
| 目的 观察基于不同参数重建光子计数CT预测胃癌T分期的价值。方法 回顾性收集14例接受腹部增强光子计数CT扫描的胃癌患者，分别基于不同卷积核（BR40、BR48、BR56）+不同等级量子迭代重建（QIR）技术（Q2、Q3、Q4）组合方案进行重建，共获得9组CT图像；以病理学结果为金标准，对比观察上述图像用于胃癌T分期的效能。结果BR40及BR48重建图像中，基于Q2及Q3评估胃癌T分期的准确率均为85.71%（12/14）、Q4达100%（14/14）；且基于BR40+Q4及BR48+Q4评估结果与金标准的一致性极高（κ均=1.000），而基于BR40+Q2、BR40+Q3、BR48+Q2及BR48+Q3评估结果与金标准的一致性均较高（κ均=0.763）。BR56重建图像中，基于Q2、Q3、Q4的评估准确率分别为57.14%（8/14）、35.71%（5/14）及42.86%（6/14），与金标准的一致性均差（κ=0.354、0.149、0.211）。9种重建图像之间，信噪比（SNR）和对比度噪声比（CNR）差异均有统计学意义（P均＜0.001），均以BR40+Q4方案较佳。结论 基于BR40+Q4重建光子计数CT用于预测胃癌T分期效能较佳。 |
| 英文摘要: |
| Objective To observe the value of photon-counting CT based on different reconstruction parameters for predicting T-stage of gastric cancer. Methods Totally 14 patients with gastric cancer who underwent abdominal enhanced photon-counting CT scanning were retrospectively enrolled. CT images were reconstructed with different combinations of different convolution kernels，i.e. （BR40， BR48， BR56）+different quantum iterative reconstruction （QIR） techniques （Q2， Q3， Q4）， and 9 sets of images were finally obtained. Taken pathological results as the gold standards， the efficacy of the above images for predicting T-staging of gastric cancer was observed and compared. Results Among BR40 and BR48 reconstructed images， the accuracy based on Q2 and Q3 for predicting T-stage of gastric cancer were both 85.71% （12/14）， of Q4 were both 100% （14/14）， and the consistency of BR40+Q4 and BR48+Q4 with gold standards were both extremely high （both κ=1.000）， while of BR40+Q2， BR40+Q3， BR48+Q2， BR48+Q3 with the gold standards were all relatively high （all κ=0.763）. Among BR56 reconstructed images， the accuracy based on Q2， Q3 and Q4 was 57.14% （8/14）， 35.71% （5/14） and 42.86% （6/14）， respectively， all had poor consistency with the gold standards （κ=0.354， 0.149， 0.211）. Significant differences of signal-to-noise ratio （SNR） and contrast-to-noise ratio （CNR） were found among 9 sets of images （both P＜0.001）， and BR40+Q4 was the best. Conclusion Photon-counting reconstructed CT images based on BR40+Q4 had better efficacy for predicting T-stage of gastric cancer. |
| 查看全文 查看/发表评论 下载PDF阅读器 |
| 关闭 |
