Attention-Guided Curriculum Learning for Weakly Supervised Classification and Localization of Thoracic Diseases on Chest Radiographs View Full Text


Ontology type: schema:Chapter      Open Access: True


Chapter Info

DATE

2018

AUTHORS

Yuxing Tang , Xiaosong Wang , Adam P. Harrison , Le Lu , Jing Xiao , Ronald M. Summers

ABSTRACT

In this work, we exploit the task of joint classification and weakly supervised localization of thoracic diseases from chest radiographs, with only image-level disease labels coupled with disease severity-level (DSL) information of a subset. A convolutional neural network (CNN) based attention-guided curriculum learning (AGCL) framework is presented, which leverages the severity-level attributes mined from radiology reports. Images in order of difficulty (grouped by different severity-levels) are fed to CNN to boost the learning gradually. In addition, highly confident samples (measured by classification probabilities) and their corresponding class-conditional heatmaps (generated by the CNN) are extracted and further fed into the AGCL framework to guide the learning of more distinctive convolutional features in the next iteration. A two-path network architecture is designed to regress the heatmaps from selected seed samples in addition to the original classification task. The joint learning scheme can improve the classification and localization performance along with more seed samples for the next iteration. We demonstrate the effectiveness of this iterative refinement framework via extensive experimental evaluations on the publicly available ChestXray14 dataset. AGCL achieves over 5.7% (averaged over 14 diseases) increase in classification AUC and 7%/11% increases in Recall/Precision for the localization task compared to the state of the art. More... »

PAGES

249-258

References to SciGraph publications

Book

TITLE

Machine Learning in Medical Imaging

ISBN

978-3-030-00918-2
978-3-030-00919-9

Identifiers

URI

http://scigraph.springernature.com/pub.10.1007/978-3-030-00919-9_29

DOI

http://dx.doi.org/10.1007/978-3-030-00919-9_29

DIMENSIONS

https://app.dimensions.ai/details/publication/pub.1107037353


Indexing Status Check whether this publication has been indexed by Scopus and Web Of Science using the SN Indexing Status Tool
Incoming Citations Browse incoming citations for this publication using opencitations.net

JSON-LD is the canonical representation for SciGraph data.

TIP: You can open this SciGraph record using an external JSON-LD service: JSON-LD Playground Google SDTT

[
  {
    "@context": "https://springernature.github.io/scigraph/jsonld/sgcontext.json", 
    "about": [
      {
        "id": "http://purl.org/au-research/vocabulary/anzsrc-for/2008/0801", 
        "inDefinedTermSet": "http://purl.org/au-research/vocabulary/anzsrc-for/2008/", 
        "name": "Artificial Intelligence and Image Processing", 
        "type": "DefinedTerm"
      }, 
      {
        "id": "http://purl.org/au-research/vocabulary/anzsrc-for/2008/08", 
        "inDefinedTermSet": "http://purl.org/au-research/vocabulary/anzsrc-for/2008/", 
        "name": "Information and Computing Sciences", 
        "type": "DefinedTerm"
      }
    ], 
    "author": [
      {
        "affiliation": {
          "alternateName": "National Institutes of Health Clinical Center", 
          "id": "https://www.grid.ac/institutes/grid.410305.3", 
          "name": [
            "National Institutes of Health, Clinical Center"
          ], 
          "type": "Organization"
        }, 
        "familyName": "Tang", 
        "givenName": "Yuxing", 
        "id": "sg:person.014425570327.67", 
        "sameAs": [
          "https://app.dimensions.ai/discover/publication?and_facet_researcher=ur.014425570327.67"
        ], 
        "type": "Person"
      }, 
      {
        "affiliation": {
          "alternateName": "National Institutes of Health Clinical Center", 
          "id": "https://www.grid.ac/institutes/grid.410305.3", 
          "name": [
            "National Institutes of Health, Clinical Center"
          ], 
          "type": "Organization"
        }, 
        "familyName": "Wang", 
        "givenName": "Xiaosong", 
        "id": "sg:person.012233025131.02", 
        "sameAs": [
          "https://app.dimensions.ai/discover/publication?and_facet_researcher=ur.012233025131.02"
        ], 
        "type": "Person"
      }, 
      {
        "affiliation": {
          "alternateName": "National Institutes of Health Clinical Center", 
          "id": "https://www.grid.ac/institutes/grid.410305.3", 
          "name": [
            "National Institutes of Health, Clinical Center"
          ], 
          "type": "Organization"
        }, 
        "familyName": "Harrison", 
        "givenName": "Adam P.", 
        "id": "sg:person.015650075713.17", 
        "sameAs": [
          "https://app.dimensions.ai/discover/publication?and_facet_researcher=ur.015650075713.17"
        ], 
        "type": "Person"
      }, 
      {
        "affiliation": {
          "alternateName": "National Institutes of Health Clinical Center", 
          "id": "https://www.grid.ac/institutes/grid.410305.3", 
          "name": [
            "National Institutes of Health, Clinical Center"
          ], 
          "type": "Organization"
        }, 
        "familyName": "Lu", 
        "givenName": "Le", 
        "id": "sg:person.01353423536.73", 
        "sameAs": [
          "https://app.dimensions.ai/discover/publication?and_facet_researcher=ur.01353423536.73"
        ], 
        "type": "Person"
      }, 
      {
        "affiliation": {
          "name": [
            "Ping An Technology Co., Ltd"
          ], 
          "type": "Organization"
        }, 
        "familyName": "Xiao", 
        "givenName": "Jing", 
        "id": "sg:person.016201005410.92", 
        "sameAs": [
          "https://app.dimensions.ai/discover/publication?and_facet_researcher=ur.016201005410.92"
        ], 
        "type": "Person"
      }, 
      {
        "affiliation": {
          "alternateName": "National Institutes of Health Clinical Center", 
          "id": "https://www.grid.ac/institutes/grid.410305.3", 
          "name": [
            "National Institutes of Health, Clinical Center"
          ], 
          "type": "Organization"
        }, 
        "familyName": "Summers", 
        "givenName": "Ronald M.", 
        "id": "sg:person.011331054577.30", 
        "sameAs": [
          "https://app.dimensions.ai/discover/publication?and_facet_researcher=ur.011331054577.30"
        ], 
        "type": "Person"
      }
    ], 
    "citation": [
      {
        "id": "https://doi.org/10.1145/1553374.1553380", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1012146698"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "sg:pub.10.1007/978-3-319-46454-1_7", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1021516291", 
          "https://doi.org/10.1007/978-3-319-46454-1_7"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "https://doi.org/10.1148/radiol.2017162326", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1085056202"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "https://doi.org/10.1145/3065386", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1085642448"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "https://doi.org/10.1145/3065386", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1085642448"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "https://doi.org/10.1109/cvpr.2016.319", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1093270996"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "https://doi.org/10.1109/cvpr.2016.233", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1093655772"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "https://doi.org/10.1117/1.jmi.5.3.036501", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1105723258"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "sg:pub.10.1007/978-3-030-00934-2_81", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1107025820", 
          "https://doi.org/10.1007/978-3-030-00934-2_81"
        ], 
        "type": "CreativeWork"
      }, 
      {
        "id": "sg:pub.10.1007/978-3-030-00937-3_47", 
        "sameAs": [
          "https://app.dimensions.ai/details/publication/pub.1107028411", 
          "https://doi.org/10.1007/978-3-030-00937-3_47"
        ], 
        "type": "CreativeWork"
      }
    ], 
    "datePublished": "2018", 
    "datePublishedReg": "2018-01-01", 
    "description": "In this work, we exploit the task of joint classification and weakly supervised localization of thoracic diseases from chest radiographs, with only image-level disease labels coupled with disease severity-level (DSL) information of a subset. A convolutional neural network (CNN) based attention-guided curriculum learning (AGCL) framework is presented, which leverages the severity-level attributes mined from radiology reports. Images in order of difficulty (grouped by different severity-levels) are fed to CNN to boost the learning gradually. In addition, highly confident samples (measured by classification probabilities) and their corresponding class-conditional heatmaps (generated by the CNN) are extracted and further fed into the AGCL framework to guide the learning of more distinctive convolutional features in the next iteration. A two-path network architecture is designed to regress the heatmaps from selected seed samples in addition to the original classification task. The joint learning scheme can improve the classification and localization performance along with more seed samples for the next iteration. We demonstrate the effectiveness of this iterative refinement framework via extensive experimental evaluations on the publicly available ChestXray14 dataset. AGCL achieves over 5.7% (averaged over 14 diseases) increase in classification AUC and 7%/11% increases in Recall/Precision for the localization task compared to the state of the art.", 
    "editor": [
      {
        "familyName": "Shi", 
        "givenName": "Yinghuan", 
        "type": "Person"
      }, 
      {
        "familyName": "Suk", 
        "givenName": "Heung-Il", 
        "type": "Person"
      }, 
      {
        "familyName": "Liu", 
        "givenName": "Mingxia", 
        "type": "Person"
      }
    ], 
    "genre": "chapter", 
    "id": "sg:pub.10.1007/978-3-030-00919-9_29", 
    "inLanguage": [
      "en"
    ], 
    "isAccessibleForFree": true, 
    "isPartOf": {
      "isbn": [
        "978-3-030-00918-2", 
        "978-3-030-00919-9"
      ], 
      "name": "Machine Learning in Medical Imaging", 
      "type": "Book"
    }, 
    "name": "Attention-Guided Curriculum Learning for Weakly Supervised Classification and Localization of Thoracic Diseases on Chest Radiographs", 
    "pagination": "249-258", 
    "productId": [
      {
        "name": "doi", 
        "type": "PropertyValue", 
        "value": [
          "10.1007/978-3-030-00919-9_29"
        ]
      }, 
      {
        "name": "readcube_id", 
        "type": "PropertyValue", 
        "value": [
          "222e4d842b624b0b26ce9e7e0e734bc41bff5d4cdb3cc11b3da3012cfee76c06"
        ]
      }, 
      {
        "name": "dimensions_id", 
        "type": "PropertyValue", 
        "value": [
          "pub.1107037353"
        ]
      }
    ], 
    "publisher": {
      "location": "Cham", 
      "name": "Springer International Publishing", 
      "type": "Organisation"
    }, 
    "sameAs": [
      "https://doi.org/10.1007/978-3-030-00919-9_29", 
      "https://app.dimensions.ai/details/publication/pub.1107037353"
    ], 
    "sdDataset": "chapters", 
    "sdDatePublished": "2019-04-15T23:25", 
    "sdLicense": "https://scigraph.springernature.com/explorer/license/", 
    "sdPublisher": {
      "name": "Springer Nature - SN SciGraph project", 
      "type": "Organization"
    }, 
    "sdSource": "s3://com-uberresearch-data-dimensions-target-20181106-alternative/cleanup/v134/2549eaecd7973599484d7c17b260dba0a4ecb94b/merge/v9/a6c9fde33151104705d4d7ff012ea9563521a3ce/jats-lookup/v90/0000000001_0000000264/records_8695_00000526.jsonl", 
    "type": "Chapter", 
    "url": "http://link.springer.com/10.1007/978-3-030-00919-9_29"
  }
]
 

Download the RDF metadata as:  json-ld nt turtle xml License info

HOW TO GET THIS DATA PROGRAMMATICALLY:

JSON-LD is a popular format for linked data which is fully compatible with JSON.

curl -H 'Accept: application/ld+json' 'https://scigraph.springernature.com/pub.10.1007/978-3-030-00919-9_29'

N-Triples is a line-based linked data format ideal for batch operations.

curl -H 'Accept: application/n-triples' 'https://scigraph.springernature.com/pub.10.1007/978-3-030-00919-9_29'

Turtle is a human-readable linked data format.

curl -H 'Accept: text/turtle' 'https://scigraph.springernature.com/pub.10.1007/978-3-030-00919-9_29'

RDF/XML is a standard XML format for linked data.

curl -H 'Accept: application/rdf+xml' 'https://scigraph.springernature.com/pub.10.1007/978-3-030-00919-9_29'


 

This table displays all metadata directly associated to this object as RDF triples.

142 TRIPLES      23 PREDICATES      36 URIs      20 LITERALS      8 BLANK NODES

Subject Predicate Object
1 sg:pub.10.1007/978-3-030-00919-9_29 schema:about anzsrc-for:08
2 anzsrc-for:0801
3 schema:author Nd7cd84ff68894de9b688816fbfae0b8f
4 schema:citation sg:pub.10.1007/978-3-030-00934-2_81
5 sg:pub.10.1007/978-3-030-00937-3_47
6 sg:pub.10.1007/978-3-319-46454-1_7
7 https://doi.org/10.1109/cvpr.2016.233
8 https://doi.org/10.1109/cvpr.2016.319
9 https://doi.org/10.1117/1.jmi.5.3.036501
10 https://doi.org/10.1145/1553374.1553380
11 https://doi.org/10.1145/3065386
12 https://doi.org/10.1148/radiol.2017162326
13 schema:datePublished 2018
14 schema:datePublishedReg 2018-01-01
15 schema:description In this work, we exploit the task of joint classification and weakly supervised localization of thoracic diseases from chest radiographs, with only image-level disease labels coupled with disease severity-level (DSL) information of a subset. A convolutional neural network (CNN) based attention-guided curriculum learning (AGCL) framework is presented, which leverages the severity-level attributes mined from radiology reports. Images in order of difficulty (grouped by different severity-levels) are fed to CNN to boost the learning gradually. In addition, highly confident samples (measured by classification probabilities) and their corresponding class-conditional heatmaps (generated by the CNN) are extracted and further fed into the AGCL framework to guide the learning of more distinctive convolutional features in the next iteration. A two-path network architecture is designed to regress the heatmaps from selected seed samples in addition to the original classification task. The joint learning scheme can improve the classification and localization performance along with more seed samples for the next iteration. We demonstrate the effectiveness of this iterative refinement framework via extensive experimental evaluations on the publicly available ChestXray14 dataset. AGCL achieves over 5.7% (averaged over 14 diseases) increase in classification AUC and 7%/11% increases in Recall/Precision for the localization task compared to the state of the art.
16 schema:editor N17f301a321b3419b9d7eafb23e739745
17 schema:genre chapter
18 schema:inLanguage en
19 schema:isAccessibleForFree true
20 schema:isPartOf N82a1e487cf3143fdbbed7fb94cc9ad1b
21 schema:name Attention-Guided Curriculum Learning for Weakly Supervised Classification and Localization of Thoracic Diseases on Chest Radiographs
22 schema:pagination 249-258
23 schema:productId N48711d84d80142b9afca760dc2ac97c3
24 Ncbf781128208442d966bdf7556f2f6fa
25 Ne0cf1324e0ad4ade95ff2b7860c3f63f
26 schema:publisher N280f3d4fae2345fc98019512a21f96f2
27 schema:sameAs https://app.dimensions.ai/details/publication/pub.1107037353
28 https://doi.org/10.1007/978-3-030-00919-9_29
29 schema:sdDatePublished 2019-04-15T23:25
30 schema:sdLicense https://scigraph.springernature.com/explorer/license/
31 schema:sdPublisher Nb8dc75cff8944b7d91777461c6419abd
32 schema:url http://link.springer.com/10.1007/978-3-030-00919-9_29
33 sgo:license sg:explorer/license/
34 sgo:sdDataset chapters
35 rdf:type schema:Chapter
36 N09516690b10b4e4c8f968e87e17a05a9 schema:familyName Shi
37 schema:givenName Yinghuan
38 rdf:type schema:Person
39 N17f301a321b3419b9d7eafb23e739745 rdf:first N09516690b10b4e4c8f968e87e17a05a9
40 rdf:rest Nc54109a685784a9e8a58c2904c0556e4
41 N280f3d4fae2345fc98019512a21f96f2 schema:location Cham
42 schema:name Springer International Publishing
43 rdf:type schema:Organisation
44 N48711d84d80142b9afca760dc2ac97c3 schema:name doi
45 schema:value 10.1007/978-3-030-00919-9_29
46 rdf:type schema:PropertyValue
47 N4c3a47e47856472aa39b6d56c4449aa3 schema:familyName Liu
48 schema:givenName Mingxia
49 rdf:type schema:Person
50 N6aa06fc51dda4eafafd2b12d0c73a43c rdf:first sg:person.011331054577.30
51 rdf:rest rdf:nil
52 N7291f808fcdc46c7b4f1adbf2f72f698 rdf:first sg:person.01353423536.73
53 rdf:rest Nfc0761d005904b148b28bd5143639fd3
54 N82a1e487cf3143fdbbed7fb94cc9ad1b schema:isbn 978-3-030-00918-2
55 978-3-030-00919-9
56 schema:name Machine Learning in Medical Imaging
57 rdf:type schema:Book
58 N880334b96c524074b1bab71f699acc32 rdf:first N4c3a47e47856472aa39b6d56c4449aa3
59 rdf:rest rdf:nil
60 Naa6db5e1cf5e4a128d72e34342349288 schema:familyName Suk
61 schema:givenName Heung-Il
62 rdf:type schema:Person
63 Nb8dc75cff8944b7d91777461c6419abd schema:name Springer Nature - SN SciGraph project
64 rdf:type schema:Organization
65 Nc54109a685784a9e8a58c2904c0556e4 rdf:first Naa6db5e1cf5e4a128d72e34342349288
66 rdf:rest N880334b96c524074b1bab71f699acc32
67 Ncbf781128208442d966bdf7556f2f6fa schema:name readcube_id
68 schema:value 222e4d842b624b0b26ce9e7e0e734bc41bff5d4cdb3cc11b3da3012cfee76c06
69 rdf:type schema:PropertyValue
70 Nd7cd84ff68894de9b688816fbfae0b8f rdf:first sg:person.014425570327.67
71 rdf:rest Nda5086977ace4ccdbafa292671be8bfe
72 Nda5086977ace4ccdbafa292671be8bfe rdf:first sg:person.012233025131.02
73 rdf:rest Ndcc6c250772648c5870ef8e0fc711a2c
74 Ndcc6c250772648c5870ef8e0fc711a2c rdf:first sg:person.015650075713.17
75 rdf:rest N7291f808fcdc46c7b4f1adbf2f72f698
76 Ne0cf1324e0ad4ade95ff2b7860c3f63f schema:name dimensions_id
77 schema:value pub.1107037353
78 rdf:type schema:PropertyValue
79 Nf93ff025033c4aeebfe064ab536a84c5 schema:name Ping An Technology Co., Ltd
80 rdf:type schema:Organization
81 Nfc0761d005904b148b28bd5143639fd3 rdf:first sg:person.016201005410.92
82 rdf:rest N6aa06fc51dda4eafafd2b12d0c73a43c
83 anzsrc-for:08 schema:inDefinedTermSet anzsrc-for:
84 schema:name Information and Computing Sciences
85 rdf:type schema:DefinedTerm
86 anzsrc-for:0801 schema:inDefinedTermSet anzsrc-for:
87 schema:name Artificial Intelligence and Image Processing
88 rdf:type schema:DefinedTerm
89 sg:person.011331054577.30 schema:affiliation https://www.grid.ac/institutes/grid.410305.3
90 schema:familyName Summers
91 schema:givenName Ronald M.
92 schema:sameAs https://app.dimensions.ai/discover/publication?and_facet_researcher=ur.011331054577.30
93 rdf:type schema:Person
94 sg:person.012233025131.02 schema:affiliation https://www.grid.ac/institutes/grid.410305.3
95 schema:familyName Wang
96 schema:givenName Xiaosong
97 schema:sameAs https://app.dimensions.ai/discover/publication?and_facet_researcher=ur.012233025131.02
98 rdf:type schema:Person
99 sg:person.01353423536.73 schema:affiliation https://www.grid.ac/institutes/grid.410305.3
100 schema:familyName Lu
101 schema:givenName Le
102 schema:sameAs https://app.dimensions.ai/discover/publication?and_facet_researcher=ur.01353423536.73
103 rdf:type schema:Person
104 sg:person.014425570327.67 schema:affiliation https://www.grid.ac/institutes/grid.410305.3
105 schema:familyName Tang
106 schema:givenName Yuxing
107 schema:sameAs https://app.dimensions.ai/discover/publication?and_facet_researcher=ur.014425570327.67
108 rdf:type schema:Person
109 sg:person.015650075713.17 schema:affiliation https://www.grid.ac/institutes/grid.410305.3
110 schema:familyName Harrison
111 schema:givenName Adam P.
112 schema:sameAs https://app.dimensions.ai/discover/publication?and_facet_researcher=ur.015650075713.17
113 rdf:type schema:Person
114 sg:person.016201005410.92 schema:affiliation Nf93ff025033c4aeebfe064ab536a84c5
115 schema:familyName Xiao
116 schema:givenName Jing
117 schema:sameAs https://app.dimensions.ai/discover/publication?and_facet_researcher=ur.016201005410.92
118 rdf:type schema:Person
119 sg:pub.10.1007/978-3-030-00934-2_81 schema:sameAs https://app.dimensions.ai/details/publication/pub.1107025820
120 https://doi.org/10.1007/978-3-030-00934-2_81
121 rdf:type schema:CreativeWork
122 sg:pub.10.1007/978-3-030-00937-3_47 schema:sameAs https://app.dimensions.ai/details/publication/pub.1107028411
123 https://doi.org/10.1007/978-3-030-00937-3_47
124 rdf:type schema:CreativeWork
125 sg:pub.10.1007/978-3-319-46454-1_7 schema:sameAs https://app.dimensions.ai/details/publication/pub.1021516291
126 https://doi.org/10.1007/978-3-319-46454-1_7
127 rdf:type schema:CreativeWork
128 https://doi.org/10.1109/cvpr.2016.233 schema:sameAs https://app.dimensions.ai/details/publication/pub.1093655772
129 rdf:type schema:CreativeWork
130 https://doi.org/10.1109/cvpr.2016.319 schema:sameAs https://app.dimensions.ai/details/publication/pub.1093270996
131 rdf:type schema:CreativeWork
132 https://doi.org/10.1117/1.jmi.5.3.036501 schema:sameAs https://app.dimensions.ai/details/publication/pub.1105723258
133 rdf:type schema:CreativeWork
134 https://doi.org/10.1145/1553374.1553380 schema:sameAs https://app.dimensions.ai/details/publication/pub.1012146698
135 rdf:type schema:CreativeWork
136 https://doi.org/10.1145/3065386 schema:sameAs https://app.dimensions.ai/details/publication/pub.1085642448
137 rdf:type schema:CreativeWork
138 https://doi.org/10.1148/radiol.2017162326 schema:sameAs https://app.dimensions.ai/details/publication/pub.1085056202
139 rdf:type schema:CreativeWork
140 https://www.grid.ac/institutes/grid.410305.3 schema:alternateName National Institutes of Health Clinical Center
141 schema:name National Institutes of Health, Clinical Center
142 rdf:type schema:Organization
 




Preview window. Press ESC to close (or click here)


...