| 王玉双,孙胜军,傅璠,李莹莹.脑部CT平扫图像直方图分析鉴别肿瘤性与非肿瘤性脑出血[J].中国医学影像技术,2020,36(12):1786~1789 |
| 脑部CT平扫图像直方图分析鉴别肿瘤性与非肿瘤性脑出血 |
| Histogram analysis of brain CT plain images in differential diagnosis of tumorous and non-tumorous cerebral hemorrhage |
| 投稿时间：2019-10-22 修订日期：2020-05-20 |
| DOI：10.13929/j.issn.1003-3289.2020.12.006 |
| 中文关键词: 脑出血 体层摄影术,X线计算机 直方图 鉴别诊断 |
| 英文关键词:cerebral hemorrhage tomography, X-ray computed histogram differential diagnosis |
| 基金项目:首都卫生发展科研专项项目（2018-2-1074）。 |
|
| 摘要点击次数: 2388 |
| 全文下载次数: 706 |
| 中文摘要: |
| 目的 观察脑部CT平扫图像直方图分析鉴别诊断肿瘤性与非肿瘤性脑出血的价值。方法 收集116例非肿瘤性脑出血（非肿瘤组）和44例肿瘤性脑出血（肿瘤组），分析其脑部CT平扫图像直方图，获得病灶的直方图参数，包括最小值，第5、25、50、75、95百分位CT值、最大值、标准差、平均值、偏度和峰度。比较2组间直方图参数的差异，以ROC曲线分析各参数鉴别诊断肿瘤性与非肿瘤性脑出血的效能。结果 肿瘤组最大值、第5、25、50、75、95百分位CT值、平均值、标准差均小于非肿瘤组（P均<0.05），偏度和峰度大于非肿瘤组（P均<0.05），最小值与非肿瘤组差异无统计学意义（P>0.05）。ROC曲线分析结果显示，最大值、第25、50、75、95百分位CT值、平均值及标准差鉴别肿瘤性与非肿瘤性脑出血的诊断价值中等，其中第50百分位CT值的AUC最高（0.82），敏感度和特异度分别为0.92和0.68。结论 脑部CT平扫图像直方图分析可作为鉴别诊断肿瘤性与非肿瘤性脑出血的辅助手段。 |
| 英文摘要: |
| Objective To explore the value of histogram analysis of brain CT plain images for differential diagnosis of tumorous and non-tumorous cerebral hemorrhage. Methods A total of 160 cerebral hemorrhage patients, including 116 with non-tumorous cerebral hemorrhage (non-tumorous group) and 44 with tumorous cerebral hemorrhage (tumorous group) were enrolled. Histogram analysis was performed on brain CT images of all patients to obtain histogram parameters of the lesions, including the minimum, the 5th, CT values of the 25th, 50th, 75th and 95th percentiles, the maximum,standard deviation, mean, skewness and kurtosis. The histogram parameters were compared between 2 groups, then ROC curve was used to analyze the diagnostic efficacy of different histogram parameters in differential diagnosis of tumorous and non-tumorous cerebral hemorrhage. Results The maximum, the 5th, 25th, 50th, 75th, 95th percentiles, mean and standard deviation in tumorous group were all lower than those in non-tumorous group (all P<0.05), while skewness and kurtosis were both higher in tumorous group than those in non-tumorous group (both P<0.05). No statistical difference of the minimum value was found between 2 groups (P>0.05). ROC curve analysis showed that the maximum, the 25th, 50th, 75th, 95th percentiles, mean and standard deviation were of medium diagnostic value for differentiating tumorous and non-tumorous cerebral hemorrhage. CT value of the 50th percentile had the highest AUC (0.82), with sensitivity 0.92 and specificity 0.68, respectively. Conclusion Histogram analysis of brain CT plain images could be used as an auxiliary method for differentiating tumorous and non-tumorous cerebral hemorrhage. |
| 查看全文 查看/发表评论 下载PDF阅读器 |
|
|
|
