Robust Landmark Detection in Volumetric Data with Efficient 3D Deep Learning View Full Text


Ontology type: schema:Chapter     


Chapter Info

DATE

2017-07-14

AUTHORS

Yefeng Zheng , David Liu , Bogdan Georgescu , Hien Nguyen , Dorin Comaniciu

ABSTRACT

Recently, deeplearninghasdemonstrated great success in computer vision with the capability to learn powerful image features from a large training set. However, most of the published work has been confined to solving 2D problems, with a few limited exceptions that treated the 3D space as a composition of 2D orthogonal planes. The challenge of 3D deep learning is due to a much larger input vector, compared to 2D, which dramatically increases the computation time and the chance of over-fitting, especially when combined with limited training samples (hundreds to thousands), typical for medical imaging applications. To address this challenge, we propose an efficient and robust deep learning algorithm capable of full 3D detection in volumetric data. A two-step approach is exploited for efficient detection. A shallow network (with one hidden layer) is used for the initial testing of all voxels to obtain a small number of promising candidates, followed by more accurate classification with a deep network. In addition, we propose two approaches, i.e., separable filter decompositionand network sparsification, to speed up the evaluation of a network. To mitigate the over-fitting issue, thereby increasing detection robustness, we extract small 3D patches from a multi-resolution image pyramid. The deeply learned image features are further combined with Haar wavelet-like features to increase the detection accuracy. The proposed method has been quantitatively evaluated for carotid artery bifurcation detection on a head-neck CT dataset from 455 patients. Compared to the state of the art, the mean error is reduced by more than half, from 5.97 mm to 2.64 mm, with a detection speed of less than 1 s/volume. More... »

PAGES

49-61

Book

TITLE

Deep Learning and Convolutional Neural Networks for Medical Image Computing

ISBN

978-3-319-42998-4
978-3-319-42999-1

Identifiers

URI

http://scigraph.springernature.com/pub.10.1007/978-3-319-42999-1_4

DOI

http://dx.doi.org/10.1007/978-3-319-42999-1_4

DIMENSIONS

https://app.dimensions.ai/details/publication/pub.1090665370


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/2008/08", 
        "inDefinedTermSet": "http://purl.org/au-research/vocabulary/anzsrc-for/2008/", 
        "name": "Information and Computing Sciences", 
        "type": "DefinedTerm"
      }, 
      {
        "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"
      }
    ], 
    "author": [
      {
        "affiliation": {
          "alternateName": "Medical Imaging Technologies, Siemens Healthcare, Princeton, NJ, USA", 
          "id": "http://www.grid.ac/institutes/grid.415886.6", 
          "name": [
            "Medical Imaging Technologies, Siemens Healthcare, Princeton, NJ, USA"
          ], 
          "type": "Organization"
        }, 
        "familyName": "Zheng", 
        "givenName": "Yefeng", 
        "id": "sg:person.0767211426.21", 
        "sameAs": [
          "https://app.dimensions.ai/discover/publication?and_facet_researcher=ur.0767211426.21"
        ], 
        "type": "Person"
      }, 
      {
        "affiliation": {
          "alternateName": "Medical Imaging Technologies, Siemens Healthcare, Princeton, NJ, USA", 
          "id": "http://www.grid.ac/institutes/grid.415886.6", 
          "name": [
            "Medical Imaging Technologies, Siemens Healthcare, Princeton, NJ, USA"
          ], 
          "type": "Organization"
        }, 
        "familyName": "Liu", 
        "givenName": "David", 
        "id": "sg:person.0673243411.05", 
        "sameAs": [
          "https://app.dimensions.ai/discover/publication?and_facet_researcher=ur.0673243411.05"
        ], 
        "type": "Person"
      }, 
      {
        "affiliation": {
          "alternateName": "Medical Imaging Technologies, Siemens Healthcare, Princeton, NJ, USA", 
          "id": "http://www.grid.ac/institutes/grid.415886.6", 
          "name": [
            "Medical Imaging Technologies, Siemens Healthcare, Princeton, NJ, USA"
          ], 
          "type": "Organization"
        }, 
        "familyName": "Georgescu", 
        "givenName": "Bogdan", 
        "id": "sg:person.0703547214.37", 
        "sameAs": [
          "https://app.dimensions.ai/discover/publication?and_facet_researcher=ur.0703547214.37"
        ], 
        "type": "Person"
      }, 
      {
        "affiliation": {
          "alternateName": "Medical Imaging Technologies, Siemens Healthcare, Princeton, NJ, USA", 
          "id": "http://www.grid.ac/institutes/grid.415886.6", 
          "name": [
            "Medical Imaging Technologies, Siemens Healthcare, Princeton, NJ, USA"
          ], 
          "type": "Organization"
        }, 
        "familyName": "Nguyen", 
        "givenName": "Hien", 
        "id": "sg:person.07507502363.19", 
        "sameAs": [
          "https://app.dimensions.ai/discover/publication?and_facet_researcher=ur.07507502363.19"
        ], 
        "type": "Person"
      }, 
      {
        "affiliation": {
          "alternateName": "Medical Imaging Technologies, Siemens Healthcare, Princeton, NJ, USA", 
          "id": "http://www.grid.ac/institutes/grid.415886.6", 
          "name": [
            "Medical Imaging Technologies, Siemens Healthcare, Princeton, NJ, USA"
          ], 
          "type": "Organization"
        }, 
        "familyName": "Comaniciu", 
        "givenName": "Dorin", 
        "id": "sg:person.01066111014.77", 
        "sameAs": [
          "https://app.dimensions.ai/discover/publication?and_facet_researcher=ur.01066111014.77"
        ], 
        "type": "Person"
      }
    ], 
    "datePublished": "2017-07-14", 
    "datePublishedReg": "2017-07-14", 
    "description": "Recently, deeplearninghasdemonstrated great success in computer vision with the capability to learn powerful image features from a large training set. However, most of the published work has been confined to solving 2D problems, with a few limited exceptions that treated the 3D space as a composition of 2D orthogonal planes. The challenge of 3D deep learning is due to a much larger input vector, compared to 2D, which dramatically increases the computation time and the chance of over-fitting, especially when combined with limited training samples (hundreds to thousands), typical for medical imaging applications. To address this challenge, we propose an efficient and robust deep learning algorithm capable of full 3D detection in volumetric data. A two-step approach is exploited for efficient detection. A shallow network (with one hidden layer) is used for the initial testing of all voxels to obtain a small number of promising candidates, followed by more accurate classification with a deep network. In addition, we propose two approaches, i.e., separable filter decompositionand network sparsification, to speed up the evaluation of a network. To mitigate the over-fitting issue, thereby increasing detection robustness, we extract small 3D patches from a multi-resolution image pyramid. The deeply learned image features are further combined with Haar wavelet-like features to increase the detection accuracy. The proposed method has been quantitatively evaluated for carotid artery bifurcation detection on a head-neck CT dataset from 455 patients. Compared to the state of the art, the mean error is reduced by more than half, from 5.97\u00a0mm to 2.64\u00a0mm, with a detection speed of less than 1\u00a0s/volume.", 
    "editor": [
      {
        "familyName": "Lu", 
        "givenName": "Le", 
        "type": "Person"
      }, 
      {
        "familyName": "Zheng", 
        "givenName": "Yefeng", 
        "type": "Person"
      }, 
      {
        "familyName": "Carneiro", 
        "givenName": "Gustavo", 
        "type": "Person"
      }, 
      {
        "familyName": "Yang", 
        "givenName": "Lin", 
        "type": "Person"
      }
    ], 
    "genre": "chapter", 
    "id": "sg:pub.10.1007/978-3-319-42999-1_4", 
    "inLanguage": "en", 
    "isAccessibleForFree": false, 
    "isPartOf": {
      "isbn": [
        "978-3-319-42998-4", 
        "978-3-319-42999-1"
      ], 
      "name": "Deep Learning and Convolutional Neural Networks for Medical Image Computing", 
      "type": "Book"
    }, 
    "keywords": [
      "deep learning", 
      "image features", 
      "multi-resolution image pyramid", 
      "wavelet-like features", 
      "head-neck CT", 
      "small 3D patches", 
      "Robust Landmark Detection", 
      "volumetric data", 
      "deep learning algorithms", 
      "powerful image features", 
      "large training sets", 
      "limited training samples", 
      "large input vectors", 
      "medical imaging applications", 
      "computer vision", 
      "network sparsification", 
      "deep networks", 
      "shallow network", 
      "landmark detection", 
      "image pyramid", 
      "learning algorithm", 
      "detection robustness", 
      "more accurate classification", 
      "detection accuracy", 
      "detection speed", 
      "computation time", 
      "training samples", 
      "bifurcation detection", 
      "input vector", 
      "fitting issue", 
      "training set", 
      "two-step approach", 
      "accurate classification", 
      "network", 
      "efficient detection", 
      "great success", 
      "imaging applications", 
      "learning", 
      "mean error", 
      "algorithm", 
      "detection", 
      "sparsification", 
      "features", 
      "orthogonal planes", 
      "robustness", 
      "challenges", 
      "vision", 
      "small number", 
      "capability", 
      "classification", 
      "accuracy", 
      "set", 
      "voxels", 
      "applications", 
      "error", 
      "art", 
      "initial testing", 
      "pyramid", 
      "data", 
      "space", 
      "issues", 
      "speed", 
      "vector", 
      "work", 
      "method", 
      "limited exceptions", 
      "evaluation", 
      "number", 
      "time", 
      "success", 
      "testing", 
      "patches", 
      "state", 
      "candidates", 
      "chance", 
      "plane", 
      "addition", 
      "volume", 
      "CT", 
      "exception", 
      "composition", 
      "samples", 
      "half", 
      "patients", 
      "approach", 
      "problem", 
      "robust deep learning algorithm", 
      "full 3D detection", 
      "separable filter decompositionand network sparsification", 
      "filter decompositionand network sparsification", 
      "decompositionand network sparsification", 
      "Haar wavelet-like features", 
      "carotid artery bifurcation detection", 
      "artery bifurcation detection", 
      "Efficient 3D Deep Learning"
    ], 
    "name": "Robust Landmark Detection in Volumetric Data with Efficient 3D Deep Learning", 
    "pagination": "49-61", 
    "productId": [
      {
        "name": "dimensions_id", 
        "type": "PropertyValue", 
        "value": [
          "pub.1090665370"
        ]
      }, 
      {
        "name": "doi", 
        "type": "PropertyValue", 
        "value": [
          "10.1007/978-3-319-42999-1_4"
        ]
      }
    ], 
    "publisher": {
      "name": "Springer Nature", 
      "type": "Organisation"
    }, 
    "sameAs": [
      "https://doi.org/10.1007/978-3-319-42999-1_4", 
      "https://app.dimensions.ai/details/publication/pub.1090665370"
    ], 
    "sdDataset": "chapters", 
    "sdDatePublished": "2022-01-01T19:15", 
    "sdLicense": "https://scigraph.springernature.com/explorer/license/", 
    "sdPublisher": {
      "name": "Springer Nature - SN SciGraph project", 
      "type": "Organization"
    }, 
    "sdSource": "s3://com-springernature-scigraph/baseset/20220101/entities/gbq_results/chapter/chapter_253.jsonl", 
    "type": "Chapter", 
    "url": "https://doi.org/10.1007/978-3-319-42999-1_4"
  }
]
 

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.10.1007/978-3-319-42999-1_4'

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

curl -H 'Accept: application/n-triples' 'https://scigraph.springernature.com/pub.10.1007/978-3-319-42999-1_4'

Turtle is a human-readable linked data format.

curl -H 'Accept: text/turtle' 'https://scigraph.springernature.com/pub.10.1007/978-3-319-42999-1_4'

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

curl -H 'Accept: application/rdf+xml' 'https://scigraph.springernature.com/pub.10.1007/978-3-319-42999-1_4'


 

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

198 TRIPLES      23 PREDICATES      120 URIs      113 LITERALS      7 BLANK NODES

Subject Predicate Object
1 sg:pub.10.1007/978-3-319-42999-1_4 schema:about anzsrc-for:08
2 anzsrc-for:0801
3 schema:author Nbdd4951c0e37426994c13469ac602085
4 schema:datePublished 2017-07-14
5 schema:datePublishedReg 2017-07-14
6 schema:description Recently, deeplearninghasdemonstrated great success in computer vision with the capability to learn powerful image features from a large training set. However, most of the published work has been confined to solving 2D problems, with a few limited exceptions that treated the 3D space as a composition of 2D orthogonal planes. The challenge of 3D deep learning is due to a much larger input vector, compared to 2D, which dramatically increases the computation time and the chance of over-fitting, especially when combined with limited training samples (hundreds to thousands), typical for medical imaging applications. To address this challenge, we propose an efficient and robust deep learning algorithm capable of full 3D detection in volumetric data. A two-step approach is exploited for efficient detection. A shallow network (with one hidden layer) is used for the initial testing of all voxels to obtain a small number of promising candidates, followed by more accurate classification with a deep network. In addition, we propose two approaches, i.e., separable filter decompositionand network sparsification, to speed up the evaluation of a network. To mitigate the over-fitting issue, thereby increasing detection robustness, we extract small 3D patches from a multi-resolution image pyramid. The deeply learned image features are further combined with Haar wavelet-like features to increase the detection accuracy. The proposed method has been quantitatively evaluated for carotid artery bifurcation detection on a head-neck CT dataset from 455 patients. Compared to the state of the art, the mean error is reduced by more than half, from 5.97 mm to 2.64 mm, with a detection speed of less than 1 s/volume.
7 schema:editor N03e17dc4aee341a68f7faa33b618b648
8 schema:genre chapter
9 schema:inLanguage en
10 schema:isAccessibleForFree false
11 schema:isPartOf Nc6e1c3cbce2c41b698b1790d1682ad47
12 schema:keywords CT
13 Efficient 3D Deep Learning
14 Haar wavelet-like features
15 Robust Landmark Detection
16 accuracy
17 accurate classification
18 addition
19 algorithm
20 applications
21 approach
22 art
23 artery bifurcation detection
24 bifurcation detection
25 candidates
26 capability
27 carotid artery bifurcation detection
28 challenges
29 chance
30 classification
31 composition
32 computation time
33 computer vision
34 data
35 decompositionand network sparsification
36 deep learning
37 deep learning algorithms
38 deep networks
39 detection
40 detection accuracy
41 detection robustness
42 detection speed
43 efficient detection
44 error
45 evaluation
46 exception
47 features
48 filter decompositionand network sparsification
49 fitting issue
50 full 3D detection
51 great success
52 half
53 head-neck CT
54 image features
55 image pyramid
56 imaging applications
57 initial testing
58 input vector
59 issues
60 landmark detection
61 large input vectors
62 large training sets
63 learning
64 learning algorithm
65 limited exceptions
66 limited training samples
67 mean error
68 medical imaging applications
69 method
70 more accurate classification
71 multi-resolution image pyramid
72 network
73 network sparsification
74 number
75 orthogonal planes
76 patches
77 patients
78 plane
79 powerful image features
80 problem
81 pyramid
82 robust deep learning algorithm
83 robustness
84 samples
85 separable filter decompositionand network sparsification
86 set
87 shallow network
88 small 3D patches
89 small number
90 space
91 sparsification
92 speed
93 state
94 success
95 testing
96 time
97 training samples
98 training set
99 two-step approach
100 vector
101 vision
102 volume
103 volumetric data
104 voxels
105 wavelet-like features
106 work
107 schema:name Robust Landmark Detection in Volumetric Data with Efficient 3D Deep Learning
108 schema:pagination 49-61
109 schema:productId N1ab8572feb184100b5fe55a3378dae72
110 N75f134eb56d74165ac68341eb1529e91
111 schema:publisher N78b0dc8f53f04cf980c8167b1ad50aa7
112 schema:sameAs https://app.dimensions.ai/details/publication/pub.1090665370
113 https://doi.org/10.1007/978-3-319-42999-1_4
114 schema:sdDatePublished 2022-01-01T19:15
115 schema:sdLicense https://scigraph.springernature.com/explorer/license/
116 schema:sdPublisher N9207d5ac0b47494eae07902187ae3774
117 schema:url https://doi.org/10.1007/978-3-319-42999-1_4
118 sgo:license sg:explorer/license/
119 sgo:sdDataset chapters
120 rdf:type schema:Chapter
121 N03e17dc4aee341a68f7faa33b618b648 rdf:first N27600cbe60b84c4cafa0411b8d7594aa
122 rdf:rest Nf245391590dc42df909af1e0a0a6343f
123 N0c6ce2c2df8443629ef7f66518d5e783 schema:familyName Yang
124 schema:givenName Lin
125 rdf:type schema:Person
126 N1ab8572feb184100b5fe55a3378dae72 schema:name dimensions_id
127 schema:value pub.1090665370
128 rdf:type schema:PropertyValue
129 N27600cbe60b84c4cafa0411b8d7594aa schema:familyName Lu
130 schema:givenName Le
131 rdf:type schema:Person
132 N27d924967a314ef7881df1240a92aea6 schema:familyName Carneiro
133 schema:givenName Gustavo
134 rdf:type schema:Person
135 N70712bd5dc404b07bf28917c9b56d603 rdf:first sg:person.0703547214.37
136 rdf:rest N9943c5afe31a4f16b6ffc0d3f0414981
137 N75f134eb56d74165ac68341eb1529e91 schema:name doi
138 schema:value 10.1007/978-3-319-42999-1_4
139 rdf:type schema:PropertyValue
140 N78b0dc8f53f04cf980c8167b1ad50aa7 schema:name Springer Nature
141 rdf:type schema:Organisation
142 N9207d5ac0b47494eae07902187ae3774 schema:name Springer Nature - SN SciGraph project
143 rdf:type schema:Organization
144 N922af8d223b2457b9d628ddd87a06add rdf:first N0c6ce2c2df8443629ef7f66518d5e783
145 rdf:rest rdf:nil
146 N9943c5afe31a4f16b6ffc0d3f0414981 rdf:first sg:person.07507502363.19
147 rdf:rest Ne5f5aa0d6e4a476b81ccabbbc6317f27
148 Nbdd4951c0e37426994c13469ac602085 rdf:first sg:person.0767211426.21
149 rdf:rest Ne5c2c6008362440487501cb3057a4f50
150 Nc6e1c3cbce2c41b698b1790d1682ad47 schema:isbn 978-3-319-42998-4
151 978-3-319-42999-1
152 schema:name Deep Learning and Convolutional Neural Networks for Medical Image Computing
153 rdf:type schema:Book
154 Nd436ef238b5d492baa1fe3781daaf168 rdf:first N27d924967a314ef7881df1240a92aea6
155 rdf:rest N922af8d223b2457b9d628ddd87a06add
156 Ne5c2c6008362440487501cb3057a4f50 rdf:first sg:person.0673243411.05
157 rdf:rest N70712bd5dc404b07bf28917c9b56d603
158 Ne5f5aa0d6e4a476b81ccabbbc6317f27 rdf:first sg:person.01066111014.77
159 rdf:rest rdf:nil
160 Ne77a49cd174c4a7a8be18f3b6ca86b61 schema:familyName Zheng
161 schema:givenName Yefeng
162 rdf:type schema:Person
163 Nf245391590dc42df909af1e0a0a6343f rdf:first Ne77a49cd174c4a7a8be18f3b6ca86b61
164 rdf:rest Nd436ef238b5d492baa1fe3781daaf168
165 anzsrc-for:08 schema:inDefinedTermSet anzsrc-for:
166 schema:name Information and Computing Sciences
167 rdf:type schema:DefinedTerm
168 anzsrc-for:0801 schema:inDefinedTermSet anzsrc-for:
169 schema:name Artificial Intelligence and Image Processing
170 rdf:type schema:DefinedTerm
171 sg:person.01066111014.77 schema:affiliation grid-institutes:grid.415886.6
172 schema:familyName Comaniciu
173 schema:givenName Dorin
174 schema:sameAs https://app.dimensions.ai/discover/publication?and_facet_researcher=ur.01066111014.77
175 rdf:type schema:Person
176 sg:person.0673243411.05 schema:affiliation grid-institutes:grid.415886.6
177 schema:familyName Liu
178 schema:givenName David
179 schema:sameAs https://app.dimensions.ai/discover/publication?and_facet_researcher=ur.0673243411.05
180 rdf:type schema:Person
181 sg:person.0703547214.37 schema:affiliation grid-institutes:grid.415886.6
182 schema:familyName Georgescu
183 schema:givenName Bogdan
184 schema:sameAs https://app.dimensions.ai/discover/publication?and_facet_researcher=ur.0703547214.37
185 rdf:type schema:Person
186 sg:person.07507502363.19 schema:affiliation grid-institutes:grid.415886.6
187 schema:familyName Nguyen
188 schema:givenName Hien
189 schema:sameAs https://app.dimensions.ai/discover/publication?and_facet_researcher=ur.07507502363.19
190 rdf:type schema:Person
191 sg:person.0767211426.21 schema:affiliation grid-institutes:grid.415886.6
192 schema:familyName Zheng
193 schema:givenName Yefeng
194 schema:sameAs https://app.dimensions.ai/discover/publication?and_facet_researcher=ur.0767211426.21
195 rdf:type schema:Person
196 grid-institutes:grid.415886.6 schema:alternateName Medical Imaging Technologies, Siemens Healthcare, Princeton, NJ, USA
197 schema:name Medical Imaging Technologies, Siemens Healthcare, Princeton, NJ, USA
198 rdf:type schema:Organization
 




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


...