File Download

There are no files associated with this item.

  Links for fulltext
     (May Require Subscription)
Supplementary

Article: MS-Net: Multi-Site Network for Improving Prostate Segmentation with Heterogeneous MRI Data

TitleMS-Net: Multi-Site Network for Improving Prostate Segmentation with Heterogeneous MRI Data
Authors
Keywordsknowledge transfer
ulti-site learning
Prostatesegmentation
feature normalization
Issue Date2020
Citation
IEEE Transactions on Medical Imaging, 2020, v. 39, n. 9, p. 2713-2724 How to Cite?
AbstractAutomated prostate segmentation in MRI is highly demanded for computer-assisted diagnosis. Recently, a variety of deep learning methods have achieved remarkable progress in this task, usually relying on large amounts of training data. Due to the nature of scarcity for medical images, it is important to effectively aggregate data from multiple sites for robust model training, to alleviate the insufficiency of single-site samples. However, the prostate MRIs from different sites present heterogeneity due to the differences in scanners and imaging protocols, raising challenges for effective ways of aggregating multi-site data for network training. In this paper, we propose a novel multi-site network (MS-Net) for improving prostate segmentation by learning robust representations, leveraging multiple sources of data. To compensate for the inter-site heterogeneity of different MRI datasets, we develop Domain-Specific Batch Normalization layers in the network backbone, enabling the network to estimate statistics and perform feature normalization for each site separately. Considering the difficulty of capturing the shared knowledge from multiple datasets, a novel learning paradigm, i.e., Multi-site-guided Knowledge Transfer, is proposed to enhance the kernels to extract more generic representations from multi-site data. Extensive experiments on three heterogeneous prostate MRI datasets demonstrate that our MS-Net improves the performance across all datasets consistently, and outperforms state-of-the-art methods for multi-site learning.
Persistent Identifierhttp://hdl.handle.net/10722/299466
ISSN
2023 Impact Factor: 8.9
2023 SCImago Journal Rankings: 3.703
ISI Accession Number ID

 

DC FieldValueLanguage
dc.contributor.authorLiu, Quande-
dc.contributor.authorDou, Qi-
dc.contributor.authorYu, Lequan-
dc.contributor.authorHeng, Pheng Ann-
dc.date.accessioned2021-05-21T03:34:28Z-
dc.date.available2021-05-21T03:34:28Z-
dc.date.issued2020-
dc.identifier.citationIEEE Transactions on Medical Imaging, 2020, v. 39, n. 9, p. 2713-2724-
dc.identifier.issn0278-0062-
dc.identifier.urihttp://hdl.handle.net/10722/299466-
dc.description.abstractAutomated prostate segmentation in MRI is highly demanded for computer-assisted diagnosis. Recently, a variety of deep learning methods have achieved remarkable progress in this task, usually relying on large amounts of training data. Due to the nature of scarcity for medical images, it is important to effectively aggregate data from multiple sites for robust model training, to alleviate the insufficiency of single-site samples. However, the prostate MRIs from different sites present heterogeneity due to the differences in scanners and imaging protocols, raising challenges for effective ways of aggregating multi-site data for network training. In this paper, we propose a novel multi-site network (MS-Net) for improving prostate segmentation by learning robust representations, leveraging multiple sources of data. To compensate for the inter-site heterogeneity of different MRI datasets, we develop Domain-Specific Batch Normalization layers in the network backbone, enabling the network to estimate statistics and perform feature normalization for each site separately. Considering the difficulty of capturing the shared knowledge from multiple datasets, a novel learning paradigm, i.e., Multi-site-guided Knowledge Transfer, is proposed to enhance the kernels to extract more generic representations from multi-site data. Extensive experiments on three heterogeneous prostate MRI datasets demonstrate that our MS-Net improves the performance across all datasets consistently, and outperforms state-of-the-art methods for multi-site learning.-
dc.languageeng-
dc.relation.ispartofIEEE Transactions on Medical Imaging-
dc.subjectknowledge transfer-
dc.subjectulti-site learning-
dc.subjectProstatesegmentation-
dc.subjectfeature normalization-
dc.titleMS-Net: Multi-Site Network for Improving Prostate Segmentation with Heterogeneous MRI Data-
dc.typeArticle-
dc.description.naturelink_to_subscribed_fulltext-
dc.identifier.doi10.1109/TMI.2020.2974574-
dc.identifier.pmid32078543-
dc.identifier.scopuseid_2-s2.0-85090169653-
dc.identifier.volume39-
dc.identifier.issue9-
dc.identifier.spage2713-
dc.identifier.epage2724-
dc.identifier.eissn1558-254X-
dc.identifier.isiWOS:000566339800001-

Export via OAI-PMH Interface in XML Formats


OR


Export to Other Non-XML Formats