Sparse Annotation is Sufficient for Bootstrapping Dense Segmentation.

Vijay Venu Thiyagarajan, Arlo Sheridan, Kristen M Harris, Uri Manor
{"title":"Sparse Annotation is Sufficient for Bootstrapping Dense Segmentation.","authors":"Vijay Venu Thiyagarajan, Arlo Sheridan, Kristen M Harris, Uri Manor","doi":"10.1101/2024.06.14.599135","DOIUrl":null,"url":null,"abstract":"<p><p>Producing dense 3D reconstructions from biological imaging data is a challenging instance segmentation task that requires significant ground-truth training data for effective and accurate deep learning-based models. Generating training data requires intense human effort to annotate each instance of an object across serial section images. Our focus is on the especially complicated brain neuropil, comprising an extensive interdigitation of dendritic, axonal, and glial processes visualized through serial section electron microscopy. We developed a novel deep learning-based method to generate dense 3D segmentations rapidly from sparse 2D annotations of a few objects on single sections. Models trained on the rapidly generated segmentations achieved similar accuracy as those trained on expert dense ground-truth annotations. Human time to generate annotations was reduced by three orders of magnitude and could be produced by non-expert annotators. This capability will democratize generation of training data for large image volumes needed to achieve brain circuits and measures of circuit strengths.</p>","PeriodicalId":72407,"journal":{"name":"bioRxiv : the preprint server for biology","volume":null,"pages":null},"PeriodicalIF":0.0000,"publicationDate":"2024-10-26","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11195258/pdf/","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"bioRxiv : the preprint server for biology","FirstCategoryId":"1085","ListUrlMain":"https://doi.org/10.1101/2024.06.14.599135","RegionNum":0,"RegionCategory":null,"ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"","JCRName":"","Score":null,"Total":0}
引用次数: 0

Abstract

Producing dense 3D reconstructions from biological imaging data is a challenging instance segmentation task that requires significant ground-truth training data for effective and accurate deep learning-based models. Generating training data requires intense human effort to annotate each instance of an object across serial section images. Our focus is on the especially complicated brain neuropil, comprising an extensive interdigitation of dendritic, axonal, and glial processes visualized through serial section electron microscopy. We developed a novel deep learning-based method to generate dense 3D segmentations rapidly from sparse 2D annotations of a few objects on single sections. Models trained on the rapidly generated segmentations achieved similar accuracy as those trained on expert dense ground-truth annotations. Human time to generate annotations was reduced by three orders of magnitude and could be produced by non-expert annotators. This capability will democratize generation of training data for large image volumes needed to achieve brain circuits and measures of circuit strengths.

查看原文
分享 分享
微信好友 朋友圈 QQ好友 复制链接
本刊更多论文
基于深度学习的策略,从稀疏注释的二维图像中生成密集的三维分割。
从生物成像数据中生成密集的三维重建是一项具有挑战性的实例分割任务,需要大量的地面实况训练数据,才能建立有效、准确的基于深度学习的模型。生成训练数据需要大量的人力来注释序列切片图像中的每个对象实例。我们的研究重点是特别复杂的脑神经髓质,它包括通过序列切片电子显微镜观察到的树突、轴突和神经胶质过程的广泛交叉。我们开发了一种基于深度学习的新方法,可从单个切片上少数对象的稀疏二维注释中快速生成密集的三维分割。根据快速生成的分割结果训练的模型与根据专家高密度地面实况注释训练的模型具有相似的准确性。人工生成注释的时间减少了三个数量级,非专业注释人员也能生成注释。这种能力将使生成大脑回路和回路强度测量所需的大量图像的训练数据平民化。
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 去求助
来源期刊
自引率
0.00%
发文量
0
期刊最新文献
Discovery of a multipotent cell type from the term human placenta. Phenotypic complexities of rare heterozygous neurexin-1 deletions. Regulation of NRF2 by Phosphoinositides and Small Heat Shock Proteins. Mediator Subunit Med4 Enforces Metastatic Dormancy in Breast Cancer. Entropy Changes in Water Networks Promote Protein Denaturation.
×
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
现在去查看 取消
×
提示
确定
0
微信
客服QQ
Book学术公众号 扫码关注我们
反馈
×
意见反馈
请填写您的意见或建议
请填写您的手机或邮箱
已复制链接
已复制链接
快去分享给好友吧!
我知道了
×
扫码分享
扫码分享
Book学术官方微信
Book学术文献互助
Book学术文献互助群
群 号:481959085
Book学术
文献互助 智能选刊 最新文献 互助须知 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 Book学术 All rights reserved.
ghs 京公网安备 11010802042870号 京ICP备2023020795号-1