基于SLIC和GVF Snake算法的乳腺肿瘤分割 
下载： 914次
Breast Tumor Segmentation Based on SLIC and GVF Snake Algorithm
兰州交通大学电子与信息工程学院, 甘肃 兰州 730070
图 & 表
图 1. 截取乳腺MRI图像结果图
Fig. 1. Intercepting the MRI image of the breast
下载图片 查看原文
图 2. SLIC结果对比图。(a)原始图像;(b)预处理结果图;(c) K'<K时SLIC算法分割结果图;(d)自适应K值对应的结果图;(e) K'>K时的结果图
Fig. 2. SLIC results comparison chart. (a) Original image; (b) pre-processing result map; (c) SLIC algorithm segmentation result map when K'<K; (d) adaptive K value corresponding result map; (e) result graph when K'>K
下载图片 查看原文
图 3. 不同K值对应的PR曲线图
Fig. 3. PR curves corresponding to different K values
下载图片 查看原文
图 4. 实验结果图。(a1)~(d1)原始图像;(a2)~(d2)梯度图像;(a3)~(d3)分水岭算法结果图;(a4)~(d4)水平集算法结果图;(a5)~(d5) GVF Snake算法结果图;(a6)~(d6)本文算法结果图;(a7)~(d7)标准分割结果图
Fig. 4. Experimental results. (a1)-(d1) Original images; (a2)-(d2) gradient images; (a3)-(d3) results of watershed algorithm; (a4)-(d4) results of level set algorithm; (a5)-(d5) results of GVF Snake algorithm; (a6)-(d6) results of proposed algorithm; (a7)-(d7) results of standard segmentation
下载图片 查看原文
图 5. 不同分割模型的PR曲线和DICE线性图。(a) PR曲线图;(b) DICE线形图
Fig. 5. PR curves and DICE linear graph of different segmentation models. (a) PR curves; (b) DICE linear graph
下载图片 查看原文
表 1不同K值的分割结果评价
Table1. Evaluation of segmentation results of different K values
| Index | Fig. 1(a) | Fig. 1(b) | Fig. 1(c) |
|---|
| 100 | 125 | 150 | 50 | 71 | 100 | 15 | 18 | 50 |
|---|
| DICE | 0.8307 | 0.8419 | 0.8242 | 0.8382 | 0.8441 | 0.8301 | 0.7657 | 0.7700 | 0.7394 | | VOE | 0.3383 | 0.3160 | 0.3515 | 0.3236 | 0.3117 | 0.3397 | 0.4686 | 0.4599 | 0.5212 |
|
查看原文
表 2不同K值对应的平均绝对误差
Table2. Average absolute error corresponding to different K values
| Index | K'>K | K'<K | K'=K |
|---|
| MAE | 0.173802 | 0.172418 | 0.171266 |
|
查看原文
表 3不同分割模型的评价指标
Table3. Evaluation indicators for different segmentation models
| Algorithm index | DICE | VOE | RVD | Precision | Recall |
|---|
| Fig. 4(a) | Fig. 4(b) | Fig. 4(a) | Fig. 4(b) | Fig. 4(a) | Fig. 4(b) | Fig. 4(a) | Fig. 4(b) | Fig. 4(a) | Fig. 4(b) |
|---|
| Level set | 0.9101 | 0.9173 | 0.1662 | 0.1621 | 0.1810 | 0.1763 | 0.8403 | 0.8485 | 0.9925 | 0.9981 | | SLIC | 0.8419 | 0.8441 | 0.3160 | 0.3117 | 0.3752 | 0.3693 | 0.7271 | 0.7303 | 0.9204 | 0.9327 | | GVF Snake | 0.8987 | 0.8898 | 0.2024 | 0.1018 | 0.2252 | 0.0018 | 0.8161 | 0.7870 | 0.9258 | 0.9483 | | Watershed | 0.9161 | 0.9122 | 0.1534 | 0.1705 | 0.1661 | 0.1864 | 0.8509 | 0.8406 | 0.9922 | 0.9973 | | SLIC+GVF Snake | 0.9305 | 0.9357 | 0.1255 | 0.0935 | 0.1339 | 0.0893 | 0.8755 | 0.9816 | 0.9928 | 0.8939 |
|
查看原文
王燕, 李积英, 杨宜林, 俞永乾, 王景慧. 基于SLIC和GVF Snake算法的乳腺肿瘤分割[J]. 激光与光电子学进展, 2020, 57(14): 141023. Yan Wang, Jiying Li, Yilin Yang, Yongqian Yu, Jinghui Wang. Breast Tumor Segmentation Based on SLIC and GVF Snake Algorithm[J]. Laser & Optoelectronics Progress, 2020, 57(14): 141023.
PDF全文
