Fetal Cardiac Structure Detection Using Multi-task Learning

被引:0
作者
He, Jie [1 ,2 ]
Yang, Lei [1 ,3 ]
Zhu, Yunping [5 ,6 ,7 ]
Li, Donglian [4 ]
Ding, Zhixing [8 ]
Lu, Yuhuan [1 ]
Liang, Bocheng [9 ]
Li, Shengli [9 ]
机构
[1] Hunan Univ, Coll Comp Sci & Elect Engn, Changsha 410000, Hunan, Peoples R China
[2] Wuzhou Univ, Guangxi Key Lab Machine Vis & Intelligent Control, Wuzhou 543000, Guangxi, Peoples R China
[3] Hunan Univ Chongqing, Res Inst, Chongqing 400000, Peoples R China
[4] Guilin Univ Elect Technol, Sch Informat & Commun, Guilin 541004, Guangxi, Peoples R China
[5] Natl Ctr Prot Sci, Beijing 102206, Peoples R China
[6] Beijing Proteome Res Ctr, Beijing 102206, Peoples R China
[7] Beijing Inst Life, Beijing 102206, Peoples R China
[8] Yunnan Univ, Sch Informat Sci & Engn, Kunming 650504, Yunnan, Peoples R China
[9] Shenzhen Maternal & Child Healthcare Hosp, Dept Ultrasound, Shenzhen 518028, Guangdong, Peoples R China
来源
ADVANCED INTELLIGENT COMPUTING IN BIOINFORMATICS, PT II, ICIC 2024 | 2024年 / 14882卷
基金
中国国家自然科学基金;
关键词
Ultrasound standard plane; Multi-task learning; Anatomical structure detection; Feature fusion;
D O I
10.1007/978-981-97-5692-6_36
中图分类号
TP18 [人工智能理论];
学科分类号
081104 ; 0812 ; 0835 ; 1405 ;
摘要
The prenatal ultrasound examination is a highly effective method for evaluating fetal health, wherein the acquisition of standard fetal ultrasound planes and accurate identification of key structures within them are essential for precise measurement of fetal growth parameters. However, manual acquisition of standard planes is not only time-consuming and laborious, but may also introduce subjectivity and inconsistency due to individual experience differences among sonographers. To mitigate these challenges, we propose a fetal cardiac structure detection model based on multi-task learning (FCSD), which adopts the ConvNeXt network structure to extract underlying shared multi-scale features. Within the detection architecture, the Feature Pyramid Network (FPN) integrates disparate feature layers from ConvNeXt, enabling the comprehensive utilization of semantic information encapsulated in high-level feature maps, thereby enhancing the capability to extract crucial features. Additionally, we design a classification module with stacked multi-layer residual networks and feature fusion, which enables efficient and accurate discrimination of standard planes in fetal ultrasound images. The experimental outcomes show that the proposed FCSD model surpasses other competitive baseline models in terms of detection accuracy and classification precision, which can assist sonographers in prenatal screening.
引用
收藏
页码:405 / 419
页数:15
相关论文
共 35 条
[1]  
Carion N., 2020, LNCS, V12346, P213, DOI [DOI 10.1007/978-3-030-58452-813, 10.1007/978- 3- 030-58452-8 13, DOI 10.1007/978-3-030-58452-8_13]
[2]   MLMSeg: A multi-view learning model for ultrasound thyroid nodule segmentation [J].
Chen, Guanyuan ;
Tan, Guanghua ;
Duan, Mingxing ;
Pu, Bin ;
Luo, Hongxia ;
Li, Shengli ;
Li, Kenli .
COMPUTERS IN BIOLOGY AND MEDICINE, 2024, 169
[3]   You Only Look One-level Feature [J].
Chen, Qiang ;
Wang, Yingming ;
Yang, Tong ;
Zhang, Xiangyu ;
Cheng, Jian ;
Sun, Jian .
2021 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION, CVPR 2021, 2021, :13034-13043
[4]  
Dosovitskiy A, 2021, Arxiv, DOI arXiv:2010.11929
[5]  
Howard AG, 2017, Arxiv, DOI arXiv:1704.04861
[6]  
Ge Z, 2021, Arxiv, DOI [arXiv:2107.08430, 10.48550/arXiv.2107.08430, DOI 10.48550/ARXIV.2107.08430]
[7]   Fast R-CNN [J].
Girshick, Ross .
2015 IEEE INTERNATIONAL CONFERENCE ON COMPUTER VISION (ICCV), 2015, :1440-1448
[8]   Rich feature hierarchies for accurate object detection and semantic segmentation [J].
Girshick, Ross ;
Donahue, Jeff ;
Darrell, Trevor ;
Malik, Jitendra .
2014 IEEE CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR), 2014, :580-587
[9]   VP-Nets : Efficient automatic localization of key brain structures in 3D fetal neurosonography [J].
Huang, Ruobing ;
Xie, Weidi ;
Noble, J. Alison .
MEDICAL IMAGE ANALYSIS, 2018, 47 :127-139
[10]  
Liang B., 2024, Neurocomputing