| 王薇,张冀.基于主成分分析模型的肝双模超声造影改良呼吸运动校正法[J].中国医学影像技术,2022,38(2):280~285 |
| 基于主成分分析模型的肝双模超声造影改良呼吸运动校正法 |
| Improved respiratory motion correction method based on principal components analysis model for dual-mode contrast-enhanced ultrasound of liver |
| 投稿时间：2021-06-08 修订日期：2021-11-22 |
| DOI：10.13929/j.issn.1003-3289.2022.02.027 |
| 中文关键词: 肝肿瘤 主成分分析 呼吸运动 运动校正 超声检查 |
| 英文关键词:liver neoplasms principal component analysis respiratory motion motion correction ultrasonography |
| 基金项目: |
|
| 摘要点击次数: 1924 |
| 全文下载次数: 417 |
| 中文摘要: |
| 目的 提出基于主成分分析（PCA）模型的改良呼吸运动校正法用于肝双模超声造影（CEUS）图像，并观察其应用价值。方法 纳入11例肝内肿瘤患者，采集肝肿瘤CEUS图像，将所获肝内肿瘤CEUS视频转换为图像序列，基于灰阶子图像序列建立PCA模型，以其主成分合成最佳呼吸运动曲线；选取CEUS峰期段与呼吸曲线均值相位处最近的图像作为参照图，并于呼吸周期间隔内选取与参照图相位处最近的图像作为校正后的灰阶图像序列；观察校正前、后图像质量，分析改良呼吸运动校正法的价值。结果 校正后，从CEUS图像序列ROI中提取的时间-强度曲线（TIC）较校正前更为平滑，且图像序列平均图像结构相似度（MSSIM）和平均相关系数（MCC）均明显大于校正前（P均<0.05），而偏差值（DV）明显小于校正前（P<0.05）。结论 基于PCA模型的改良呼吸运动校正法无需参数即可选择参照图，用于肝双模CEUS图像序列具有一定价值。 |
| 英文摘要: |
| Objective To develop the improved respiratory motion correction method based on principal component analysis (PCA) model for dual-mode contrast-enhanced ultrasound (CEUS), and to observe its value for liver CEUS. Methods Eleven patients with liver tumors were enrolled, and CEUS images were collected. CEUS of liver tumors were converted into image sequences, and PCA model of respiratory motion was established based on the grayscale subimage sequences, and the principal components from the data were leveraged to generate the optimal respiratory curve. Then, the image closest to the mean phase of the respiratory curve was considered as the reference image. Finally, the frames with the nearest phase positions to the reference image within the same interval of the respiratory cycle were selected for grayscale image sequence after correction. The image quality before and after correction were observed, and the value of improved respiratory motion correction method was analyzed. Results The time-intensity curves (TIC) extracted from ROI of CEUS image sequences after correction were smoother than those before correction, and the mean structural similarity (MSSIM) and mean correlation coefficient (MCC) of image sequences were both larger than those before correction (both P<0.05), while deviation value (DV) of image sequences was smaller than that before correction (P<0.05). Conclusion The improved respiratory motion correction method based on PCA model could select the reference image without parameters, which had certain value for liver dual-mode CEUS image sequences. |
| 查看全文 查看/发表评论 下载PDF阅读器 |
