《计算机应用》唯一官方网站 ›› 2024, Vol. 44 ›› Issue (4): 1285-1293.DOI: 10.11772/j.issn.1001-9081.2023040517
所属专题: 多媒体计算与计算机仿真
李威1, 陈玲2, 徐修远1, 朱敏2, 郭际香1, 周凯1, 牛颢1, 张煜宸3, 易珊烨3, 章毅1, 罗凤鸣2()
收稿日期:
2023-05-04
修回日期:
2023-07-13
接受日期:
2023-07-14
发布日期:
2023-08-03
出版日期:
2024-04-10
通讯作者:
罗凤鸣
作者简介:
李威(1994—),男,四川绵竹人,硕士研究生,CCF会员,主要研究方向:医学图像分割基金资助:
Wei LI1, Ling CHEN2, Xiuyuan XU1, Min ZHU2, Jixiang GUO1, Kai ZHOU1, Hao NIU1, Yuchen ZHANG3, Shanye YI3, Yi ZHANG1, Fengming LUO2()
Received:
2023-05-04
Revised:
2023-07-13
Accepted:
2023-07-14
Online:
2023-08-03
Published:
2024-04-10
Contact:
Fengming LUO
About author:
LI Wei, born in 1994, M. S. candidate. His research interests include medical image segmentation.Supported by:
摘要:
间质性肺病(ILD)的分割标签标注成本极高,且现有数据集通常存在样本量较少的问题,导致训练的模型效果较差。针对该问题,提出一种基于多任务学习的ILD分割算法。首先,基于U-Net构建多任务分割模型;其次,使用生成的肺部分割标签作为辅助任务标签进行多任务学习;最后,使用一种自适应调整多任务损失函数权重的方法,平衡主任务和辅助任务的损失。在自构建的ILD数据集上的实验结果表明,多任务分割模型的Dice相似系数(DSC)达到了82.61%,与U-Net相比提升了2.26个百分点。验证了所提算法可以提升ILD的分割性能,协助临床医生进行ILD诊断。
中图分类号:
李威, 陈玲, 徐修远, 朱敏, 郭际香, 周凯, 牛颢, 张煜宸, 易珊烨, 章毅, 罗凤鸣. 基于多任务学习的间质性肺病分割算法[J]. 计算机应用, 2024, 44(4): 1285-1293.
Wei LI, Ling CHEN, Xiuyuan XU, Min ZHU, Jixiang GUO, Kai ZHOU, Hao NIU, Yuchen ZHANG, Shanye YI, Yi ZHANG, Fengming LUO. Interstitial lung disease segmentation algorithm based on multi-task learning[J]. Journal of Computer Applications, 2024, 44(4): 1285-1293.
编号 | 模型 | 设置 | DSC |
---|---|---|---|
对照组1 | U-Net | 单任务 | 80.35 |
对照组2 | U-Net | 预训练+微调 | 81.18 |
对照组3 | 多任务分割模型 | 多数据集预训练+微调 | 81.90 |
对照组4 | U-Net | 肺分割过滤输入 | 79.06 |
对照组5 | 多任务分割模型 | 单任务 | 80.26 |
实验组 | 多任务分割模型 | 伪标签+多任务 | 82.61 |
表1 不同对照组的分割性能对比 (%)
Tab. 1 Comparison of segmentation performance of different groups
编号 | 模型 | 设置 | DSC |
---|---|---|---|
对照组1 | U-Net | 单任务 | 80.35 |
对照组2 | U-Net | 预训练+微调 | 81.18 |
对照组3 | 多任务分割模型 | 多数据集预训练+微调 | 81.90 |
对照组4 | U-Net | 肺分割过滤输入 | 79.06 |
对照组5 | 多任务分割模型 | 单任务 | 80.26 |
实验组 | 多任务分割模型 | 伪标签+多任务 | 82.61 |
主任务权重 | 辅助任务权重 | DSC/% |
---|---|---|
0.95 | 0.05 | 81.42 |
0.90 | 0.10 | 81.95 |
0.80 | 0.20 | 82.15 |
0.70 | 0.30 | 82.12 |
0.60 | 0.40 | 81.56 |
0.50 | 0.50 | 81.11 |
0.40 | 0.60 | 80.77 |
0.20 | 0.80 | 79.53 |
0.05 | 0.95 | 77.80 |
自适应 | 自适应 | 82.61 |
表2 多任务损失函数对比实验结果
Tab. 2 Comparison experiment results of multi-task loss functions
主任务权重 | 辅助任务权重 | DSC/% |
---|---|---|
0.95 | 0.05 | 81.42 |
0.90 | 0.10 | 81.95 |
0.80 | 0.20 | 82.15 |
0.70 | 0.30 | 82.12 |
0.60 | 0.40 | 81.56 |
0.50 | 0.50 | 81.11 |
0.40 | 0.60 | 80.77 |
0.20 | 0.80 | 79.53 |
0.05 | 0.95 | 77.80 |
自适应 | 自适应 | 82.61 |
模型 | DSC |
---|---|
SegNet[ | 76.42 |
DeepLabv3+[ | 78.95 |
PNet[ | 79.35 |
Swin-Unet[ | 72.41 |
TransUNet[ | 76.67 |
本文模型 | 82.61 |
表3 不同模型的分割性能对比 (%)
Tab. 3 Comparison of segmentation performance of different models
模型 | DSC |
---|---|
SegNet[ | 76.42 |
DeepLabv3+[ | 78.95 |
PNet[ | 79.35 |
Swin-Unet[ | 72.41 |
TransUNet[ | 76.67 |
本文模型 | 82.61 |
模型 | DSC | ||
---|---|---|---|
20%样本量 | 50%样本量 | 100%样本量 | |
PNet[ | 75.78 | 78.69 | 79.35 |
本文模型 | 80.04 | 81.24 | 82.61 |
表4 不同样本量的分割性能对比 (%)
Tab. 4 Comparison of segmentation performance with different sample sizes
模型 | DSC | ||
---|---|---|---|
20%样本量 | 50%样本量 | 100%样本量 | |
PNet[ | 75.78 | 78.69 | 79.35 |
本文模型 | 80.04 | 81.24 | 82.61 |
1 | CHOI Y, LIU T T, PANKRATZ D G, et al. Identification of usual interstitial pneumonia pattern using RNA-Seq and machine learning: challenges and solutions[J]. BMC Genomics, 2018, 19(): 101. 10.1186/s12864-018-4467-6 |
2 | British Thoracic Society, Standards of Care Committee. The diagnosis, assessment and treatment of diffuse parenchymal lung disease in adults[J]. Thorax, 1999, 54(): S1-S14. 10.1136/thx.54.suppl_1.s1 |
3 | 何渝娇,李风森.间质性肺病治疗的研究进展[J].新疆中医药,2019,37(6):102-105. |
HE Y J, LI F S. Research progress in the treatment of interstitial lung disease [J]. Xinjiang Journal of Traditional Chinese Medicine, 2019, 37(6): 102-105. | |
4 | WIDELL J, LIDÉN M. Interobserver variability in high-resolution CT of the lungs[J]. European Journal of Radiology Open, 2020, 7: 100228. 10.1016/j.ejro.2020.100228 |
5 | 施俊,汪琳琳,王珊珊,等.深度学习在医学影像中的应用综述[J].中国图象图形学报,2020,25(10):1953-1981. 10.11834/jig.200255 |
SHI J, WANG L L, WANG S S, et al. Applications of deep learning in medical imaging: a survey [J]. Journal of Image and Graphics, 2020, 25(10): 1953-1981. 10.11834/jig.200255 | |
6 | 曹玉红,徐海,刘荪傲,等.基于深度学习的医学影像分割研究综述[J].计算机应用,2021,41(8):2273-2287. 10.11772/j.issn.1001-9081.2020101638 |
CAO Y H, XU H, LIU S A, et al. Review of deep learning-based medical image segmentation [J]. Journal of Computer Applications, 2021, 41(8): 2273-2287. 10.11772/j.issn.1001-9081.2020101638 | |
7 | AGARWALA S, KALE M, KUMAR D, et al. Deep learning for screening of interstitial lung disease patterns in high-resolution CT images[J]. Clinical Radiology, 2020, 75(6): 481. 10.1016/j.crad.2020.01.010 |
8 | BUONGIORNO R, GERMANESE D, ROMEI C, et al. UIP-Net: a decoder-encoder CNN for the detection and quantification of usual interstitial pneumoniae pattern in lung CT scan images[C]// Proceedings of the 2021 International Conference on Pattern Recognition. Cham: Springer, 2021: 389-405. 10.1007/978-3-030-68763-2_30 |
9 | VAKALOPOULOU M, CHASSAGNON G, PARAGIOS N, et al. Deep patch-based priors under a fully convolutional encoder-decoder architecture for interstitial lung disease segmentation[C]// Proceedings of the 2018 IEEE 15th International Symposium on Biomedical Imaging. Piscataway: IEEE, 2018: 855-858. 10.1109/isbi.2018.8363706 |
10 | DEPEURSINGE A, VARGAS A, PLATON A, et al. Building a reference multimedia database for interstitial lung diseases[J]. Computerized Medical Imaging and Graphics, 2012, 36(3): 227-238. 10.1016/j.compmedimag.2011.07.003 |
11 | RUDER S. An overview of multi-task learning in deep neural networks [EB/OL]. (2017-06-15) [2023-05-01]. . |
12 | RONNEBERGER O, FISCHER P, BROX T. U-Net: convolutional networks for biomedical image segmentation[C]// Proceedings of the 18th International Conference on Medical Image Computing and Computer-Assisted Intervention. Cham: Springer, 2015: 234-241. 10.1007/978-3-319-24574-4_28 |
13 | ZHANG Y, YANG Q. A survey on multi-task learning[J]. IEEE Transactions on Knowledge and Data Engineering, 2022, 34(12): 5586-5609. 10.1109/tkde.2021.3070203 |
14 | CIPOLLA R, GAL Y, KENDALL A. Multi-task learning using uncertainty to weigh losses for scene geometry and semantics[C]// Proceedings of the 2018 IEEE Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2018: 7482-7491. 10.1109/cvpr.2018.00781 |
15 | HARRISON A P, XU Z, GEORGE K, et al. Progressive and multi-path holistically nested neural networks for pathological lung segmentation from CT images[C]// Proceedings of the 20th International Conference on Medical Image Computing and Computer Assisted Intervention. Cham: Springer, 2017: 621-629. 10.1007/978-3-319-66179-7_71 |
16 | KHANNA A, LONDHE N D, GUPTA S, et al. A deep residual U-Net convolutional neural network for automated lung segmentation in computed tomography images[J]. Biocybernetics and Biomedical Engineering, 2020, 40(3): 1314-1327. 10.1016/j.bbe.2020.07.007 |
17 | LIU C, ZHAO R, PANG M. Lung segmentation based on random forest and multi-scale edge detection[J]. IET Image Processing, 2019, 13(10): 1745-1754. 10.1049/iet-ipr.2019.0130 |
18 | AGARWALA S, KUMAR A, DHARA A K, et al. Special convolutional neural network for identification and positioning of interstitial lung disease patterns in computed tomography images[J]. Pattern Recognition and Image Analysis, 2021, 31: 730-738. 10.1134/s1054661821040027 |
19 | THORAT O, SALVI S, DEDHIA S, et al. Domain adaptation and weight initialization of neural networks for diagnosing interstitial lung diseases[J]. International Journal of Imaging Systems and Technology, 2022, 32(5): 1535-1547. 10.1002/ima.22716 |
20 | JUNAYED M S, JENY A A, ISLAM M B, et al. An efficient end-to-end deep neural network for interstitial lung disease recognition and classification[J]. Turkish Journal of Electrical Engineering and Computer Sciences, 2022, 30(4): 1235-1250. 10.55730/1300-0632.3846 |
21 | ANTHIMOPOULOS M, CHRISTODOULIDIS S, EBNER L, et al. Semantic segmentation of pathological lung tissue with dilated fully convolutional networks[J]. IEEE journal of Biomedical and Health Informatics, 2019, 23(2): 714-722. 10.1109/jbhi.2018.2818620 |
22 | LIAO F, LIANG M, LI Z, et al. Evaluate the malignancy of pulmonary nodules using the 3-D deep leaky noisy-OR network[J]. IEEE Transactions on Neural Networks and Learning Systems, 2019, 30(11): 3484-3495. 10.1109/tnnls.2019.2892409 |
23 | XU X, WANG C, GUO J, et al. DeepLN: a framework for automatic lung nodule detection using multi-resolution CT screening images[J]. Knowledge-Based Systems, 2020, 189: 105128. 10.1016/j.knosys.2019.105128 |
24 | KINGMA D P, BA J. Adam: a method for stochastic optimization [EB/OL]. (2017-01-30) [2023-05-01]. . |
25 | BADRINARAYANAN V, KENDALL A, CIPOLLA R. SegNet: a deep convolutional encoder-decoder architecture for image segmentation[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2017, 39(12): 2481-2495. 10.1109/tpami.2016.2644615 |
26 | CHEN L-C, ZHU Y, PAPANDREOU G, et al. Encoder-decoder with atrous separable convolution for semantic image segmentation[C]// Proceedings of the 2018 European Conference on Computer Vision. Cham: Springer, 2018: 833-851. 10.1007/978-3-030-01234-2_49 |
27 | CHEN J, LU Y, YU Q, et al. TransUNet: Transformers make strong encoders for medical image segmentation[EB/OL]. (2021-02-08) [2023-05-01]. . 10.48550/arXiv.2102.04306 |
28 | CAO H, WANG Y, CHEN J, et al. Swin-Unet: Unet-like pure transformer for medical image segmentation[C]// Proceedings of the 2022 European Conference on Computer Vision. Cham: Springer, 2023: 205-218. 10.1007/978-3-031-25066-8_9 |
29 | LONG J, SHELHAMER E, DARRELL T. Fully convolutional networks for semantic segmentation[C]// Proceedings of the 2015 IEEE Conference on Computer Vision and Pattern Recognition. Piscataway: IEEE, 2015: 3431-3440. 10.1109/cvpr.2015.7298965 |
[1] | 王铂越, 李英祥, 钟剑丹. 基于改进Res-UNet的昼夜地基云图分割网络[J]. 《计算机应用》唯一官方网站, 2024, 44(4): 1310-1316. |
[2] | 张鹏飞, 韩李涛, 冯恒健, 李洪梅. 基于注意力机制和全局特征优化的点云语义分割[J]. 《计算机应用》唯一官方网站, 2024, 44(4): 1086-1092. |
[3] | 尚爱国, 朱欣娟. 基于多任务学习的意图检测和槽位填充联合方法[J]. 《计算机应用》唯一官方网站, 2024, 44(3): 690-695. |
[4] | 吴宁, 罗杨洋, 许华杰. 基于多尺度特征融合的遥感图像语义分割方法[J]. 《计算机应用》唯一官方网站, 2024, 44(3): 737-744. |
[5] | 宋钰丹, 王晶, 王雪徽, 马朝阳, 林友芳. 基于自适应多任务学习的睡眠生理时序分类方法[J]. 《计算机应用》唯一官方网站, 2024, 44(2): 654-662. |
[6] | 廖存燚, 郑毅, 刘玮瑾, 于欢, 刘守印. 自动驾驶环境感知多任务去耦-融合算法[J]. 《计算机应用》唯一官方网站, 2024, 44(2): 424-431. |
[7] | 李子怡, 曲婷婷, 崇乾鹏, 徐金东. 基于模糊多尺度特征的遥感图像分割网络[J]. 《计算机应用》唯一官方网站, 2024, 44(11): 3581-3586. |
[8] | 刘永江, 陈斌. 基于多尺度记忆库的像素级无监督工业异常检测[J]. 《计算机应用》唯一官方网站, 2024, 44(11): 3587-3594. |
[9] | 郑秋梅, 牛薇薇, 王风华, 赵丹. 基于细节增强的双分支实时语义分割网络[J]. 《计算机应用》唯一官方网站, 2024, 44(10): 3058-3066. |
[10] | 郭晓, 陈艳平, 唐瑞雪, 黄瑞章, 秦永彬. 融合行为词的罪名预测多任务学习模型[J]. 《计算机应用》唯一官方网站, 2024, 44(1): 159-166. |
[11] | 周迪, 张自力, 陈佳, 胡新荣, 何儒汉, 张俊. 基于EfficientNetV2和物体上下文表示的胃癌图像分割方法[J]. 《计算机应用》唯一官方网站, 2023, 43(9): 2955-2962. |
[12] | 郑帅, 张晓龙, 邓鹤, 任宏伟. 基于多尺度特征融合和网格注意力机制的三维肝脏影像分割方法[J]. 《计算机应用》唯一官方网站, 2023, 43(7): 2303-2310. |
[13] | 鲁斌, 柳杰林. 基于特征增强的三维点云语义分割[J]. 《计算机应用》唯一官方网站, 2023, 43(6): 1818-1825. |
[14] | 何建辉, 胡春龙, 束鑫. 基于多峰标签分布学习的多任务年龄估计方法[J]. 《计算机应用》唯一官方网站, 2023, 43(5): 1578-1583. |
[15] | 袁泉, 徐雲鹏, 唐成亮. 基于路径标签的文档级关系抽取方法[J]. 《计算机应用》唯一官方网站, 2023, 43(4): 1029-1035. |
阅读次数 | ||||||
全文 |
|
|||||
摘要 |
|
|||||