| 王旭,刘义军,赵明月,李贝贝,范勇,童小雨,王诗耕.重建矩阵对CT影像组学特征稳定性的影响:体模研究[J].中国医学影像技术,2022,38(9):1412~1416 |
| 重建矩阵对CT影像组学特征稳定性的影响:体模研究 |
| Impact of reconstruction matrix size on stability of CT radiomics features:A phantom study |
| 投稿时间：2022-03-04 修订日期：2022-06-13 |
| DOI：10.13929/j.issn.1003-3289.2022.09.030 |
| 中文关键词: 体层摄影术,X线计算机 体模,显像术 影像组学 重建矩阵 |
| 英文关键词:tomography, X-ray computed phantoms, imaging radiomics reconstruction matrix |
| 基金项目: |
|
| 摘要点击次数: 1390 |
| 全文下载次数: 361 |
| 中文摘要: |
| 目的 分析不同大小重建矩阵对CT影像组学特征稳定性的影响。方法 向PH-75体模中依次置入软组织棒(肝脏等价材料)及4、8、12 mgI/ml共 3种浓度的碘棒；采用uCT 760对体模进行扫描，扫描矩阵512×512，滤波反投影(FBP)算法，卷积核(B_SOFT_B，软组织)。将数据传输至uCT 760图像后处理工作站，保持扫描野与显示野一致，基于原始数据分别重建矩阵为512×512、768×768及1 024×1 024的3组图像；以ITK-SNAP软件手动勾画体模中软组织及碘棒的ROI，采用AK软件提取其7类共计107个影像组学特征，计算各特征变异系数(CV)，评价特征稳定性:CV ≤ 5%为非常稳定，5%20%为不稳定。结果 重建矩阵改变对部分一阶(first order)及灰度游程矩阵(GLRLM)特征影响较大(CV>20%，红色)，而对形状(shape)、灰度共生矩阵(GLCM)、灰度区域矩阵(GLSZM)、灰度相关矩阵(GLDM)及邻域灰度差分矩阵(NGTDM)特征影响较小(CV ≤ 10%)。重建矩阵分别为512×512、768×768及1 024×1 024时，软组织分别有102、95及96个特征非常稳定； 4 mgI/ml碘棒分别有84、86及99个，8 mgI/ml碘棒分别有87、95及99个，12 mgI/ml碘棒分别有89、101及104个特征非常稳定。结论 重建矩阵影响CT影像组学特征的稳定性，且其程度与组织成分及强化程度有关。 |
| 英文摘要: |
| Objective To explore the impact of different reconstruction matrix sizes on the stability of CT radiomics features. Methods Soft tissue rod (liver equivalent material) and iodine rods (4, 8 and 12 mgI/ml) were inserted into a PH-75 phantom, which was then scanned using uCT 760 with scanning matrix 512×512, filtered back projection (FBP) algorithm and convolution kernel (B_SOFT_B, soft tissue). The data were transferred to the uCT 760 image post-processing workstation, and 3 groups of images were obtained with 512×512, 768×768 and 1 024×1 024 reconstruction matrices, respectively, based on the original data under the scanning field consistent with the display field. ITK-SNAP software was used to manually sketch ROI of soft tissue and iodine rods in the phantom, while AK software was used to extract a total of 107 features of 7 categories of radiomics for each ROI. Coefficient of variation (CV) of each feature was calculated, and the feature's stability was evaluated. CV ≤ 5% was considered as very stable, 5%20% was considered as unstable. Results Changing of reconstruction matrix had great impact on the characteristics of partial first order and gray level run length matrix (GLRLM) (CV>20%, red), but limited impact on the shape, gray level co-occurrence matrix (GLCM), gray level size zone matrix (GLSZM), gray level dependence matrix (GLDM) and neighborhood gray-tone difference matrix (NGTDM) (CV ≤ 10%). Under 512×512, 768×768 and 1 024×1 024 reconstruction matrices, there were 102, 95 and 96 very stable features of soft tissue, 84, 86 and 99 very stable features of 4 mgI/ml iodine rod, 87, 95 and 99 very stable features of 8 mgI/ml iodine rod, 89, 101 and 104 very stable features of 12 mgI/ml iodine rod, respectively. Conclusion The changes in reconstruction matrix affected the stability of CT radiomics features, and the degree was related to tissue composition and enhancement degree. |
| 查看全文 查看/发表评论 下载PDF阅读器 |
|
|
|
