Mamba- and ResNet-Based Dual-Branch Network for Ultrasound Thyroid Nodule Segmentation.

IF 3.8 3区 医学 Q2 ENGINEERING, BIOMEDICAL
Min Hu, Yaorong Zhang, Huijun Xue, Hao Lv, Shipeng Han
{"title":"Mamba- and ResNet-Based Dual-Branch Network for Ultrasound Thyroid Nodule Segmentation.","authors":"Min Hu, Yaorong Zhang, Huijun Xue, Hao Lv, Shipeng Han","doi":"10.3390/bioengineering11101047","DOIUrl":null,"url":null,"abstract":"<p><p>Accurate segmentation of thyroid nodules in ultrasound images is crucial for the diagnosis of thyroid cancer and preoperative planning. However, the segmentation of thyroid nodules is challenging due to their irregular shape, blurred boundary, and uneven echo texture. To address these challenges, a novel Mamba- and ResNet-based dual-branch network (MRDB) is proposed. Specifically, the visual state space block (VSSB) from Mamba and ResNet-34 are utilized to construct a dual encoder for extracting global semantics and local details, and establishing multi-dimensional feature connections. Meanwhile, an upsampling-convolution strategy is employed in the left decoder focusing on image size and detail reconstruction. A convolution-upsampling strategy is used in the right decoder to emphasize gradual feature refinement and recovery. To facilitate the interaction between local details and global context within the encoder and decoder, cross-skip connection is introduced. Additionally, a novel hybrid loss function is proposed to improve the boundary segmentation performance of thyroid nodules. Experimental results show that MRDB outperforms the state-of-the-art approaches with DSC of 90.02% and 80.6% on two public thyroid nodule datasets, TN3K and TNUI-2021, respectively. Furthermore, experiments on a third external dataset, DDTI, demonstrate that our method improves the DSC by 10.8% compared to baseline and exhibits good generalization to clinical small-scale thyroid nodule datasets. The proposed MRDB can effectively improve thyroid nodule segmentation accuracy and has great potential for clinical applications.</p>","PeriodicalId":8874,"journal":{"name":"Bioengineering","volume":null,"pages":null},"PeriodicalIF":3.8000,"publicationDate":"2024-10-20","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11504408/pdf/","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Bioengineering","FirstCategoryId":"5","ListUrlMain":"https://doi.org/10.3390/bioengineering11101047","RegionNum":3,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q2","JCRName":"ENGINEERING, BIOMEDICAL","Score":null,"Total":0}
引用次数: 0

Abstract

Accurate segmentation of thyroid nodules in ultrasound images is crucial for the diagnosis of thyroid cancer and preoperative planning. However, the segmentation of thyroid nodules is challenging due to their irregular shape, blurred boundary, and uneven echo texture. To address these challenges, a novel Mamba- and ResNet-based dual-branch network (MRDB) is proposed. Specifically, the visual state space block (VSSB) from Mamba and ResNet-34 are utilized to construct a dual encoder for extracting global semantics and local details, and establishing multi-dimensional feature connections. Meanwhile, an upsampling-convolution strategy is employed in the left decoder focusing on image size and detail reconstruction. A convolution-upsampling strategy is used in the right decoder to emphasize gradual feature refinement and recovery. To facilitate the interaction between local details and global context within the encoder and decoder, cross-skip connection is introduced. Additionally, a novel hybrid loss function is proposed to improve the boundary segmentation performance of thyroid nodules. Experimental results show that MRDB outperforms the state-of-the-art approaches with DSC of 90.02% and 80.6% on two public thyroid nodule datasets, TN3K and TNUI-2021, respectively. Furthermore, experiments on a third external dataset, DDTI, demonstrate that our method improves the DSC by 10.8% compared to baseline and exhibits good generalization to clinical small-scale thyroid nodule datasets. The proposed MRDB can effectively improve thyroid nodule segmentation accuracy and has great potential for clinical applications.

基于 Mamba 和 ResNet 的双分支网络用于超声甲状腺结节分类
准确分割超声图像中的甲状腺结节对于诊断甲状腺癌和制定术前计划至关重要。然而,由于甲状腺结节形状不规则、边界模糊、回声纹理不均匀,因此对其进行分割具有挑战性。为了应对这些挑战,我们提出了一种基于 Mamba 和 ResNet 的新型双分支网络 (MRDB)。具体来说,利用 Mamba 和 ResNet-34 的视觉状态空间块(VSSB)构建双编码器,以提取全局语义和局部细节,并建立多维特征连接。同时,左解码器采用了上采样-卷积策略,重点关注图像大小和细节重建。右侧解码器采用卷积-上采样策略,强调逐步完善和恢复特征。为了促进编码器和解码器内局部细节与全局背景之间的互动,引入了交叉跳转连接。此外,还提出了一种新的混合损失函数,以提高甲状腺结节的边界分割性能。实验结果表明,在 TN3K 和 TNUI-2021 这两个公开甲状腺结节数据集上,MRDB 的 DSC 分别为 90.02% 和 80.6%,优于最先进的方法。此外,在第三个外部数据集 DDTI 上的实验表明,与基线相比,我们的方法将 DSC 提高了 10.8%,并在临床小规模甲状腺结节数据集上表现出良好的泛化能力。所提出的 MRDB 能有效提高甲状腺结节分割的准确性,在临床应用中具有很大的潜力。
本文章由计算机程序翻译,如有差异,请以英文原文为准。
求助全文
约1分钟内获得全文 求助全文
来源期刊
Bioengineering
Bioengineering Chemical Engineering-Bioengineering
CiteScore
4.00
自引率
8.70%
发文量
661
期刊介绍: Aims Bioengineering (ISSN 2306-5354) provides an advanced forum for the science and technology of bioengineering. It publishes original research papers, comprehensive reviews, communications and case reports. Our aim is to encourage scientists to publish their experimental and theoretical results in as much detail as possible. All aspects of bioengineering are welcomed from theoretical concepts to education and applications. There is no restriction on the length of the papers. The full experimental details must be provided so that the results can be reproduced. There are, in addition, four key features of this Journal: ● We are introducing a new concept in scientific and technical publications “The Translational Case Report in Bioengineering”. It is a descriptive explanatory analysis of a transformative or translational event. Understanding that the goal of bioengineering scholarship is to advance towards a transformative or clinical solution to an identified transformative/clinical need, the translational case report is used to explore causation in order to find underlying principles that may guide other similar transformative/translational undertakings. ● Manuscripts regarding research proposals and research ideas will be particularly welcomed. ● Electronic files and software regarding the full details of the calculation and experimental procedure, if unable to be published in a normal way, can be deposited as supplementary material. ● We also accept manuscripts communicating to a broader audience with regard to research projects financed with public funds. Scope ● Bionics and biological cybernetics: implantology; bio–abio interfaces ● Bioelectronics: wearable electronics; implantable electronics; “more than Moore” electronics; bioelectronics devices ● Bioprocess and biosystems engineering and applications: bioprocess design; biocatalysis; bioseparation and bioreactors; bioinformatics; bioenergy; etc. ● Biomolecular, cellular and tissue engineering and applications: tissue engineering; chromosome engineering; embryo engineering; cellular, molecular and synthetic biology; metabolic engineering; bio-nanotechnology; micro/nano technologies; genetic engineering; transgenic technology ● Biomedical engineering and applications: biomechatronics; biomedical electronics; biomechanics; biomaterials; biomimetics; biomedical diagnostics; biomedical therapy; biomedical devices; sensors and circuits; biomedical imaging and medical information systems; implants and regenerative medicine; neurotechnology; clinical engineering; rehabilitation engineering ● Biochemical engineering and applications: metabolic pathway engineering; modeling and simulation ● Translational bioengineering
×
引用
GB/T 7714-2015
复制
MLA
复制
APA
复制
导出至
BibTeX EndNote RefMan NoteFirst NoteExpress
×
提示
您的信息不完整,为了账户安全,请先补充。
现在去补充
×
提示
您因"违规操作"
具体请查看互助需知
我知道了
×
提示
确定
请完成安全验证×
copy
已复制链接
快去分享给好友吧!
我知道了
右上角分享
点击右上角分享
0
联系我们:info@booksci.cn Book学术提供免费学术资源搜索服务,方便国内外学者检索中英文文献。致力于提供最便捷和优质的服务体验。 Copyright © 2023 布克学术 All rights reserved.
京ICP备2023020795号-1
ghs 京公网安备 11010802042870号
Book学术文献互助
Book学术文献互助群
群 号:481959085
Book学术官方微信