现代仪器与医疗

基于多专家系统的长尾内窥镜视频伪影检测算法研究

PDF下载

金浩龙 , 冯庆敏 , 刘胜林 ^*

现代仪器与医疗 | 医学工程 2025,31(3): 10-16

收起

现代仪器与医疗 |医学工程 2025 , 31 (3) : 10 -16

基于多专家系统的长尾内窥镜视频伪影检测算法研究

全屏

[Author(id=1320388942168441476, tenantId=1146029695717560320, journalId=1319373289681944729, articleId=1320388939572167293, orderNo=0, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=null, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1320388942264910470, tenantId=1146029695717560320, journalId=1319373289681944729, articleId=1320388939572167293, authorId=1320388942168441476, language=EN, stringName=Haolong Jin, firstName=Haolong, middleName=null, lastName=Jin, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=null, address=Institute of Biomedical Engineering, Union Hospital, Tongji Medical College, Huazhong University of Science and Technology, Wuhan 430000, China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1320388942369768071, tenantId=1146029695717560320, journalId=1319373289681944729, articleId=1320388939572167293, authorId=1320388942168441476, language=CN, stringName=金浩龙, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=null, address=华中科技大学同济医学院附属协和医院生物医学工程研究所，武汉　430000, bio={"content":"

金浩龙，男，在读硕士研究生，研究方向：医学图像处理、长尾学习

"}, bioImg=null, bioContent=

金浩龙，男，在读硕士研究生，研究方向：医学图像处理、长尾学习

, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1320388942067778176, tenantId=1146029695717560320, journalId=1319373289681944729, articleId=1320388939572167293, xref=null, ext=[AuthorCompanyExt(id=1320388942080361089, tenantId=1146029695717560320, journalId=1319373289681944729, articleId=1320388939572167293, companyId=1320388942067778176, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=Institute of Biomedical Engineering, Union Hospital, Tongji Medical College, Huazhong University of Science and Technology, Wuhan 430000, China), AuthorCompanyExt(id=1320388942092944002, tenantId=1146029695717560320, journalId=1319373289681944729, articleId=1320388939572167293, companyId=1320388942067778176, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=华中科技大学同济医学院附属协和医院生物医学工程研究所，武汉　430000)])]), Author(id=1320388942449459849, tenantId=1146029695717560320, journalId=1319373289681944729, articleId=1320388939572167293, orderNo=1, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=null, emailSecond=null, emailThird=null, correspondingAuthor=0, authorType=1, ext={EN=AuthorExt(id=1320388942533345931, tenantId=1146029695717560320, journalId=1319373289681944729, articleId=1320388939572167293, authorId=1320388942449459849, language=EN, stringName=Qingmin Feng, firstName=Qingmin, middleName=null, lastName=Feng, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=null, address=Institute of Biomedical Engineering, Union Hospital, Tongji Medical College, Huazhong University of Science and Technology, Wuhan 430000, China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1320388942608843404, tenantId=1146029695717560320, journalId=1319373289681944729, articleId=1320388939572167293, authorId=1320388942449459849, language=CN, stringName=冯庆敏, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=null, address=华中科技大学同济医学院附属协和医院生物医学工程研究所，武汉　430000, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1320388942067778176, tenantId=1146029695717560320, journalId=1319373289681944729, articleId=1320388939572167293, xref=null, ext=[AuthorCompanyExt(id=1320388942080361089, tenantId=1146029695717560320, journalId=1319373289681944729, articleId=1320388939572167293, companyId=1320388942067778176, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=Institute of Biomedical Engineering, Union Hospital, Tongji Medical College, Huazhong University of Science and Technology, Wuhan 430000, China), AuthorCompanyExt(id=1320388942092944002, tenantId=1146029695717560320, journalId=1319373289681944729, articleId=1320388939572167293, companyId=1320388942067778176, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=华中科技大学同济医学院附属协和医院生物医学工程研究所，武汉　430000)])]), Author(id=1320388942696923790, tenantId=1146029695717560320, journalId=1319373289681944729, articleId=1320388939572167293, orderNo=2, firstName=null, middleName=null, lastName=null, nameCn=null, orcid=null, stid=null, country=null, authorPic=null, dead=0, email=null, emailSecond=null, emailThird=null, correspondingAuthor=1, authorType=1, ext={EN=AuthorExt(id=1320388942793392784, tenantId=1146029695717560320, journalId=1319373289681944729, articleId=1320388939572167293, authorId=1320388942696923790, language=EN, stringName=Shenglin Liu, firstName=Shenglin, middleName=null, lastName=Liu, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=^*, address=Institute of Biomedical Engineering, Union Hospital, Tongji Medical College, Huazhong University of Science and Technology, Wuhan 430000, China, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null), CN=AuthorExt(id=1320388942856307345, tenantId=1146029695717560320, journalId=1319373289681944729, articleId=1320388939572167293, authorId=1320388942696923790, language=CN, stringName=刘胜林, firstName=null, middleName=null, lastName=null, prefix=null, suffix=null, authorComment=null, nameInitials=null, affiliation=null, department=null, xref=^*, address=华中科技大学同济医学院附属协和医院生物医学工程研究所，武汉　430000, bio=null, bioImg=null, bioContent=null, aboutCorrespAuthor=null)}, companyList=[AuthorCompany(id=1320388942067778176, tenantId=1146029695717560320, journalId=1319373289681944729, articleId=1320388939572167293, xref=null, ext=[AuthorCompanyExt(id=1320388942080361089, tenantId=1146029695717560320, journalId=1319373289681944729, articleId=1320388939572167293, companyId=1320388942067778176, language=EN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=Institute of Biomedical Engineering, Union Hospital, Tongji Medical College, Huazhong University of Science and Technology, Wuhan 430000, China), AuthorCompanyExt(id=1320388942092944002, tenantId=1146029695717560320, journalId=1319373289681944729, articleId=1320388939572167293, companyId=1320388942067778176, language=CN, country=null, province=null, city=null, postcode=null, companyName=null, departmentName=null, remark=华中科技大学同济医学院附属协和医院生物医学工程研究所，武汉　430000)])])]

金浩龙, 冯庆敏, 刘胜林^*

作者信息

华中科技大学同济医学院附属协和医院生物医学工程研究所，武汉　430000

通讯作者:

刘胜林，男，博士研究生，高级工程师，研究方向：人工智能医疗器械、医学图像处理、裸眼三维可视化技术

作者简介:

金浩龙，男，在读硕士研究生，研究方向：医学图像处理、长尾学习

Long-tailed endoscopic artifact detection based on multi-expert system

Haolong Jin, Qingmin Feng, Shenglin Liu^*

Affiliations

Institute of Biomedical Engineering, Union Hospital, Tongji Medical College, Huazhong University of Science and Technology, Wuhan 430000, China

出版时间: 2025-06-10 doi: 10.11876/mimt202503002

文章导航

摘要

收起

目的

通过长尾学习的视角探讨内窥镜视频伪影检测的方法。

方法

通过分析公开数据集EAD 2020中内窥镜视频伪影分布特性，采用解耦框架，提出一种基于多专家系统的长尾内窥镜伪影检测算法，将头部类别与尾部类别分为两组，不同专家处理不同组别，并且设计了专家结果聚合的方式，最后通过引入视觉提示（visual prompt）辅助专家，提高了伪影检测的全类平均精度（mean Average Precision，mAP）。

结果

得益于更强大的主干网络与更先进的分类器设计，相比于现有流行的集成方法，内窥镜连续视频帧检测的mAP从0.440提高至0.491，独立视频帧检测的mAP从0.339提升至0.384，并且本方法在所有类别上全面领先于集成方法。

结论

基于多专家系统的长尾内窥镜视频伪影检测框架通过分组与多专家有效平衡了不同伪影检测的表现，有效提高了尾部类别的检测性能，同时没有损害头部类别的性能，缓解了类别不平衡带来的问题，为数据不平衡场景下的医学图像检测任务提供了一个新的解决思路。

关键词

深度学习 / 长尾学习 / 内窥镜 / 伪影检测 / 多专家系统 / 视觉提示

Abstract

收起

Objective

Exploring the challenges of artifact detection in endoscopic videos through the perspective of long-tailed learning.

Methods

Through analysis of artifact distribution characteristics in endoscopic videos from the public EAD2020 dataset, we develop a long-tailed endoscopic artifact detection algorithm via a decoupled framework. The proposed multi-expert system strategically partitions head and tail classes into two distinct subsets, with domain-specific experts processing each group separately. We innovatively design a hierarchical expert aggregation mechanism and further integrate visual prompts to enhance cross-group discrimination capabilities, ultimately achieving a significant improvement in artifact detection performance as measured by the comprehensive mean Average Precision (mAP) metric.

Results

Benefiting from a more powerful backbone network and advanced classifier design, the proposed method achieves significant improvements in endoscopic video analysis: mAP for sequential video frame detection increases from 0.440 to 0.491, while mAP for individual video frame detection rises from 0.339 to 0.384. Notably, our approach comprehensively outperforms existing ensemble methods across all artifact categories.

Conclusion

The proposed multi-expert framework for long-tailed endoscopic video artifact detection achieves balanced performance across artifact categories through grouping and multi-expert methods. This architecture significantly enhances detection performance for tail classes while maintaining head class accuracy, effectively mitigating the inherent class imbalance problem. This study provides a novel direction for medical image detection tasks, particularly in data-imbalanced scenarios.

Key words

Deep learning / Long-tailed learning / Endoscope / Artifact detection / Multi-expert system / Visual prompt

引用本文

金浩龙, 冯庆敏, 刘胜林. 基于多专家系统的长尾内窥镜视频伪影检测算法研究. 现代仪器与医疗, 2025 , 31 (3) : 10 -16 . DOI: 10.11876/mimt202503002

Haolong Jin, Qingmin Feng, Shenglin Liu. Long-tailed endoscopic artifact detection based on multi-expert system[J]. Modern Instruments & Medical Treatment, 2025 , 31 (3) : 10 -16 . DOI: 10.11876/mimt202503002

基金

收起

湖北省重点研发计划项目(2022BCA025)

参考文献

收起

文献

收起

参考文献引证文献

排序方式：

[1]

Ali S, Zhou F, Braden B, et al. An objective comparison of detection and segmentation algorithms for artefacts in clinical endoscopy[J]. Sci Rep, 2020, 10(1): 2748.

[2]

Ali S. Where do we stand in AI for endoscopic image analysis? Deciphering gaps and future directions[J]. NPJ Digit Med, 2022, 5(1): 184.

[3]

Ali S, Zhou F, Daul C, et al. Endoscopy artifact detection (EAD 2019) challenge dataset[EB/OL]. (2019-05-08) [2025-04-14]. https://arxiv.org/abs/1905.03209.

[4]

Yang L, Jiang H, Song Q, et al. A survey on long-tailed visual recognition[J]. International Journal of Computer Vision, 2022, 130: 1837-1872.

[5]

Zhang Y, Kang B, Hooi B, et al. Deep long-tailed learning: a survey[J]. IEEE Trans Pattern Anal Mach Intell, 2023, 45(9): 10795-10816.

[6]

Yaman B, Mahmud T, Liu C H. Instance-aware repeat factor sampling for long-tailed object detection[EB/OL]. (2023-05-14) [2025-04-14]. https://arxiv.org/abs/2305.08069.

[7]

Tan J, Lu X, Zhang G, et al. Equalization loss v2: a new gradient balance approach for long-tailed object detection[C]//2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2021, Nashville, TN, USA. IEEE, 2021: 1685-1694.

[8]

Wang J, Zhang W, Zang Y, et al. Seesaw loss for long-tailed instance segmentation[C]//2021 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2021, Nashville, TN, USA. IEEE, 2021: 9690-9699.

[9]

Liu J, Sun Y, Han C, et al. Deep representation learning on long-tailed data: a learnable embedding augmentation perspective[C]//2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2020, Seattle, WA, USA. IEEE, 2020: 2967-2976.

[10]

Liu B, Li H, Kang H, et al. GistNet: a geometric structure transfer network for long-tailed recognition[C]//2021 IEEE/CVF International Conference on Computer Vision (ICCV), 2021, Montreal, QC, Canada. IEEE, 2021: 8189-8198.

[11]

He Y Y, Wu J, Wei X S. Distilling virtual examples for long-tailed recognition[C]//2021 IEEE/CVF International Conference on Computer Vision (ICCV), 2021, Montreal, QC, Canada. IEEE, 2021: 235-244.

[12]

Zhang S, Chen C, Hu X, et al. Balanced knowledge distillation for long-tailed learning[J]. Neurocomputing, 2023, 527: 36-46.

[13]

Chu P, Bian X, Liu S, et al. Feature space augmentation for long-tailed data[C]//ECCV 2020: 16th European Conference, 2020, Glasgow, UK. Springer, Cham, 2020: 694-710.

[14]

Zang Y, Huang C, Loy C C. FASA: feature augmentation and sampling adaptation for long-tailed instance segmentation[C]//2021 IEEE/CVF International Conference on Computer Vision (ICCV), 2021, Montreal, QC, Canada. IEEE, 2021: 3437-3446.

[15]

Cui J, Zhong Z, Liu S, et al. Parametric contrastive learning[C]//2021 IEEE/CVF International Conference on Computer Vision (ICCV), 2021, Montreal, QC, Canada. IEEE, 2021: 695-704.

[16]

Zhu J, Wang Z, Chen J, et al. Balanced contrastive learning for long-tailed visual recognition[C]//2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2022, New Orleans, LA, USA. IEEE, 2022: 6898-6907.

[17]

Tang K, Huang J, Zhang H. Long-tailed classification by keeping the good and removing the bad momentum causal effect[EB/OL]. (2020-09-28) [2025-04-14]. https://arxiv.org/abs/2009.12991.

[18]

Kang B, Xie S, Rohrbach M, et al. Decoupling representation and classifier for long-tailed recognition[EB/OL]. (2019-10-21) [2025-04-14]. https://arxiv.org/abs/1910.09217.

[19]

Li Y, Wang T, Kang B, et al. Overcoming classifier imbalance for long-tail object detection with balanced group softmax[C]//2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2020, Seattle, WA, USA. IEEE, 2020: 10988-10997.

[20]

Zhou B, Cui Q, Wei X S, et al. BBN: bilateral-branch network with cumulative learning for long-tailed visual recognition[C]//2020 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2020, Seattle, WA, USA. IEEE, 2020: 9716-9725.

[21]

Zhang Y, Hooi B, Hong L, et al. Self-supervised aggregation of diverse experts for test-agnostic long-tailed recognition[EB/OL]. (2021-07-20) [2025-04-14]. https://arxiv.org/abs/2107.09249.

[22]

Cai Z, Vasconcelos N. Cascade R-CNN: high quality object detection and instance segmentation[J]. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2019, 43(5): 1483-1498.

[23]

Liu Z, Mao H, Wu C Y, et al. A ConvNet for the 2020s[C]//2022 IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR), 2022, New Orleans, LA, USA. IEEE, 2022: 11966-11976.

[24]

Liu Z, Lin Y, Cao Y, et al. Swin Transformer: hierarchical vision transformer using shifted windows[C]//2021 IEEE/CVF International Conference on Computer Vision (ICCV), 2021, Montreal, QC, Canada. IEEE, 2021: 9992-10002.

[25]

Zheng L, Shen L, Tian L, et al. Scalable person reidentification: a benchmark[C]//2015 IEEE International Conference on Computer Vision (ICCV), 2015, Santiago, Chile. IEEE, 2015: 1116-1124.

2025年第31卷第3期

PDF下载

引用本文

BibTeX

文章信息

doi: 10.11876/mimt202503002

首发时间：2026-10-23
出版时间：2025-06-10

补充材料

相关文章

文章信息

作者

出版历史

基金

湖北省重点研发计划项目(2022BCA025)

作者信息

华中科技大学同济医学院附属协和医院生物医学工程研究所，武汉　430000

通讯作者:

刘胜林，男，博士研究生，高级工程师，研究方向：人工智能医疗器械、医学图像处理、裸眼三维可视化技术

参考文献

分享链接

https://castjournals.cast.org.cn/joweb/xdyqyyl/CN/10.11876/mimt202503002

分享至

全文二维码

扫描看全文

引用本文

BibTeX

本文的引用情况

2种不同金属材料的力学参数

科 Family	属数 Number of genus	种数 Number of species	占总种数比例 Percentage of total species (%)	属 Genus	种数 Number of species	占总种数比例 Percentage of total species (%)
鹅膏菌科Amanitaceae	2	11	5.26	鹅膏菌属 Amanita	10	4.78
小菇科 Mycenaceae	2	12	5.74	丝盖伞属 Inocybe	5	2.39
多孔菌科 Polyporaceae	8	14	6.70	蜡蘑属 Laccaria	5	2.39
红菇科 Russulaceae	3	23	11.00	小皮伞属 Marasmius	6	2.87
				小菇属 Mycena	11	5.26
				光柄菇属 Pluteus	5	2.39
				红菇属 Russula	17	8.13
				栓菌属 Trametes	5	2.39

关闭全屏

BibTeX
EndNote
RefWorks
TxT