In this paper, we develop a new weakly supervised\u00a0learning algorithm to learn to segment cancerous\u00a0regions in histopathology images. This paper is under a\u00a0multiple instance learning (MIL) framework with a new formulation,\u00a0deep weak supervision (DWS); we also propose\u00a0an effective way to introduce constraints to our neural\u00a0networks to assist the learning process. The contributions\u00a0of our algorithm are threefold: 1) we build an end-to-end\u00a0learning system that segments cancerous regions with fully\u00a0convolutional networks (FCNs) in which image-to-image\u00a0weakly-supervised learning is performed; 2) we develop a\u00a0DWS formulation to exploitmulti-scale learning under weak\u00a0supervision within FCNs; and 3) constraints about positive\u00a0instances are introduced in our approach to effectively\u00a0explore additional weakly supervised information that is\u00a0easy to obtain and enjoy a significant boost to the learning\u00a0process. The proposed algorithm, abbreviated as DWS-MIL,\u00a0is easy to implement and can be trained efficiently.Our system\u00a0demonstrates the state-of-the-art results on large-scale\u00a0histopathology image data sets and can be applied to various\u00a0applications inmedical imaging beyond histopathology\u00a0images, such as MRI, CT, and ultrasound images.<\/p>\n","protected":false},"excerpt":{"rendered":"
In this paper, we develop a new weakly supervised\u00a0learning algorithm to learn to segment cancerous\u00a0regions in histopathology images. This paper is under a\u00a0multiple instance learning (MIL) framework with a new formulation,\u00a0deep weak supervision (DWS); we also propose\u00a0an effective way to introduce constraints to our neural\u00a0networks to assist the learning process. The contributions\u00a0of our algorithm are […]<\/p>\n","protected":false},"featured_media":0,"template":"","meta":{"msr-url-field":"","msr-podcast-episode":"","msrModifiedDate":"","msrModifiedDateEnabled":false,"ep_exclude_from_search":false,"_classifai_error":"","footnotes":""},"msr-content-type":[3],"msr-research-highlight":[],"research-area":[13553],"msr-publication-type":[193715],"msr-product-type":[],"msr-focus-area":[],"msr-platform":[],"msr-download-source":[],"msr-locale":[268875],"msr-post-option":[],"msr-field-of-study":[],"msr-conference":[],"msr-journal":[],"msr-impact-theme":[],"msr-pillar":[],"class_list":["post-490892","msr-research-item","type-msr-research-item","status-publish","hentry","msr-research-area-medical-health-genomics","msr-locale-en_us"],"msr_publishername":"","msr_edition":"IEEE Transactions on Medical Imaging","msr_affiliation":"","msr_published_date":"2017-11-01","msr_host":"","msr_duration":"","msr_version":"","msr_speaker":"","msr_other_contributors":"","msr_booktitle":"","msr_pages_string":"","msr_chapter":"","msr_isbn":"","msr_journal":"IEEE Transactions on Medical Imaging","msr_volume":"","msr_number":"","msr_editors":"","msr_series":"","msr_issue":"","msr_organization":"","msr_how_published":"","msr_notes":"","msr_highlight_text":"","msr_release_tracker_id":"","msr_original_fields_of_study":"","msr_download_urls":"","msr_external_url":"","msr_secondary_video_url":"","msr_longbiography":"","msr_microsoftintellectualproperty":1,"msr_main_download":"490943","msr_publicationurl":"","msr_doi":"","msr_publication_uploader":[{"type":"file","title":"[2017][SCI][TMI]Constrained Deep Weak Supervision for Histopathology Image Segmentation","viewUrl":"https:\/\/www.microsoft.com\/en-us\/research\/uploads\/prod\/2018\/06\/2017SCITMIConstrained-Deep-Weak-Supervision-for-Histopathology-Image-Segmentation.pdf","id":490943,"label_id":0}],"msr_related_uploader":"","msr_attachments":[],"msr-author-ordering":[{"type":"text","value":"Zhipeng Jia","user_id":0,"rest_url":false},{"type":"text","value":"Xingyi Huang","user_id":0,"rest_url":false},{"type":"user_nicename","value":"Eric Chang","user_id":31709,"rest_url":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/microsoft-research\/v1\/researchers?person=Eric Chang"},{"type":"text","value":"Yan Xu","user_id":0,"rest_url":false}],"msr_impact_theme":[],"msr_research_lab":[199560],"msr_event":[],"msr_group":[780706],"msr_project":[170702],"publication":[],"video":[],"download":[],"msr_publication_type":"article","related_content":{"projects":[{"ID":170702,"post_title":"eHuatuo: Teaching Computer to Read Medical Records","post_name":"ehuatuo-teaching-computer-to-read-medical-records","post_type":"msr-project","post_date":"2011-04-10 20:16:13","post_modified":"2019-05-16 04:27:03","post_status":"publish","permalink":"https:\/\/www.microsoft.com\/en-us\/research\/project\/ehuatuo-teaching-computer-to-read-medical-records\/","post_excerpt":"eHuatuo is an eHealthcare project about Teaching Computer to Read Medical Records developed by Microsoft Research Asia. The goal of the project is to utilize the power of computers to help doctors process the increasing amount of data available in healthcare, ranging from text data, medical imaging data, to genomic data. We aim to link these disparate types of data together for new insights and discoveries. Resources eHuatuo: Teaching Computers to Read Medical Records Dictionary…","_links":{"self":[{"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-project\/170702"}]}}]},"_links":{"self":[{"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-research-item\/490892"}],"collection":[{"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-research-item"}],"about":[{"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/types\/msr-research-item"}],"version-history":[{"count":2,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-research-item\/490892\/revisions"}],"predecessor-version":[{"id":490976,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-research-item\/490892\/revisions\/490976"}],"wp:attachment":[{"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/media?parent=490892"}],"wp:term":[{"taxonomy":"msr-content-type","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-content-type?post=490892"},{"taxonomy":"msr-research-highlight","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-research-highlight?post=490892"},{"taxonomy":"msr-research-area","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/research-area?post=490892"},{"taxonomy":"msr-publication-type","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-publication-type?post=490892"},{"taxonomy":"msr-product-type","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-product-type?post=490892"},{"taxonomy":"msr-focus-area","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-focus-area?post=490892"},{"taxonomy":"msr-platform","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-platform?post=490892"},{"taxonomy":"msr-download-source","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-download-source?post=490892"},{"taxonomy":"msr-locale","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-locale?post=490892"},{"taxonomy":"msr-post-option","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-post-option?post=490892"},{"taxonomy":"msr-field-of-study","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-field-of-study?post=490892"},{"taxonomy":"msr-conference","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-conference?post=490892"},{"taxonomy":"msr-journal","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-journal?post=490892"},{"taxonomy":"msr-impact-theme","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-impact-theme?post=490892"},{"taxonomy":"msr-pillar","embeddable":true,"href":"https:\/\/www.microsoft.com\/en-us\/research\/wp-json\/wp\/v2\/msr-pillar?post=490892"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}