改进的体素生长算法在心脏局部血管提取中的应用 下载: 1170次
Application of Improved Voxels Growth Algorithm in Cardiac Local Vascular Extraction
1 内蒙古科技大学信息工程学院, 内蒙古 包头 014010
2 包头医学院, 内蒙古 包头 014010
图 & 表
图 1. 根据相似性查找表查找边缘体素的流程图
Fig. 1. Flow chart of finding the edge of the voxel according to similarity table
下载图片 查看原文
图 2. 血管中体素的拓扑约束示意图
Fig. 2. Topology constraint diagram of voxels in blood vessels
下载图片 查看原文
图 3. (a)胸腔三维重建及(b)局部血管簇的提取结果
Fig. 3. (a) Three-dimensional reconstruction of thoracic cavity and (b) extracted local blood vessel cluster
下载图片 查看原文
图 4. 心脏局部AABB包围盒中局部血管的提取效果
Fig. 4. Extraction of local blood vessels from cardiac local AABB bounding box
下载图片 查看原文
图 5. 算法改进前后的分割效果。(a)改进前;(b)改进后
Fig. 5. Segmentation results before and after the algorithm improving. (a) Before improving; (b) after improving
下载图片 查看原文
表 1相似性测度对比评价
Table1. Comparative evaluation for similarity measure
Similaritymeasure | Operating | Ra | Rb | Rc | Rd | Re | Adjacent voxelnumbers | Similar voxelnumbers |
---|
Meansquareerror | Beforeregistration | 1189.25 | 1147.69 | 1123.43 | 1114.16 | 1101.87 | 26 | 8 | Afterregistration | 3487.81 | 3479.15 | 3427.48 | 3404.97 | 33871.36 | 26 | 17 | Correlationcoefficient | Beforeregistration | 0.97152 | 0.97102 | 0.96785 | 0.96548 | 0.96229 | 26 | 12 | Afterregistration | 0.99241 | 0.99216 | 0.99178 | 0.99004 | 0.98459 | 26 | 18 | Mattersinformation | Beforeregistration | -1.02594 | -1.02234 | -1.02119 | -1.02101 | -1.02078 | 26 | 11 | Afterregistration | -1.37748 | -1.37536 | -1.37493 | -1.37299 | -1.37179 | 26 | 15 |
|
查看原文
表 2分割精度评价
Table2. Segmentation accuracy evaluation
Figurenumber | Traditional voxels growth algorithm | Improved voxels growth algorithm |
---|
Sensitivity | Specificity | Accuracy | Kappa | Sensitivity | Specificity | Accuracy | Kappa |
---|
1 | 0.5974 | 0.7477 | 0.8785 | 0.7664 | 0.8112 | 0.9012 | 0.9214 | 0.8785 | 2 | 0.6348 | 0.6336 | 0.8826 | 0.8204 | 0.8105 | 0.9016 | 0.9114 | 0.8568 | 3 | 0.6852 | 0.8125 | 0.8789 | 0.7344 | 0.7929 | 0.9148 | 0.9322 | 0.8897 | 4 | 0.5374 | 0.7756 | 0.8664 | 0.8010 | 0.8188 | 0.9248 | 0.9483 | 0.8924 | 5 | 0.7259 | 0.7978 | 0.8997 | 0.8452 | 0.8354 | 0.9321 | 0.9324 | 0.8899 | Averagevalue | 0.6348 | 0.8489 | 0.8347 | 0.8327 | 0.8137 | 0.9124 | 0.9247 | 0.8814 |
|
查看原文
表 3算法改进前后的体素生长评价
Table3. Voxel growth evaluation before and after algorithm improving
Algorithms andattributes | Size of boundingbox (x, y, z) | Number of voxelsin volume | Number ofvoxels on surface | Voxels growth timeinside model /s | Voxels growthtime of surface /s |
---|
Automatic voxelsgrowth algorithm | (28, 40, 23) | 26625 | 2957 | 103.89 | 5.83 | Semi-automatic voxelsgrowth algorithm | (3, 5.3, 2) | 3152 | 345 | 5.59 | 1.46 | Improved voxelsgrowth algorithm | (3, 5.3, 2) | 3152 | 345 | 3.78 | 1.12 |
|
查看原文
任国印, 吕晓琪, 杨楠, 喻大华, 张晓峰, 周涛. 改进的体素生长算法在心脏局部血管提取中的应用[J]. 激光与光电子学进展, 2018, 55(6): 061701. Guoyin Ren, Xiaoqi Lü, Nan Yang, Dahua Yu, Xiaofeng Zhang, Tao Zhou. Application of Improved Voxels Growth Algorithm in Cardiac Local Vascular Extraction[J]. Laser & Optoelectronics Progress, 2018, 55(6): 061701.