Manifold Learning for Density Segmentation in High Risk Mammograms

Harry Strange; Erika R. E. Denton; Minnie Kibiro; Reyer Zwiggelaar

doi:10.1007/978-3-642-38628-2_29

Manifold Learning for Density Segmentation in High Risk Mammograms

Harry Strange, Erika R. E. Denton, Minnie Kibiro, Reyer Zwiggelaar

Adran Cyfrifiadureg

Norfolk & Norwich University Hospital

Allbwn ymchwil: Pennod mewn Llyfr/Adroddiad/Trafodion Cynhadledd › Trafodion Cynhadledd (Nid-Cyfnodolyn fathau)

4 Dyfyniadau (Scopus)

Crynodeb

There is a strong correlation between relative mammographic breast density and the risk of developing breast cancer. As such, accurately modelling the percentage of a mammogram that is dense is a pivotal step in density based risk classification. In this work, a novel method based on manifold learning is used to segment high-risk mammograms into density regions. As such, finer details are present in the segmentations and more accurate measures of breast density are produced. A set of high risk (BI-RADS IV) full field digital mammograms with density annotations obtained from radiologists are used to test the validity of the proposed approach. By exploiting the manifold structure of the input space, segmentations with average accuracy of 87% when compared with radiologists’ segmentations can be obtained. This is an increase of over 12% compared with segmentation in the high-dimensional space.

Iaith wreiddiol	Saesneg
Teitl	Pattern Recognition and Image Analysis - 6th Iberian Conference, IbPRIA 2013, Proceedings
Is-deitl	6th Iberian Conference, IbPRIA 2013, Funchal, Madeira, Portugal, June 5-7, 2013, Proceedings
Golygyddion	João M. Sanches, Luisa Micó, Jaime S. Cardoso
Cyhoeddwr	Springer Nature
Tudalennau	245-252
Nifer y tudalennau	8
ISBN (Electronig)	978-3-642-38628-2
ISBN (Argraffiad)	978-3-642-38627-5
Dynodwyr Gwrthrych Digidol (DOIs)	https://doi.org/10.1007/978-3-642-38628-2_29
Statws	Cyhoeddwyd - 17 Mai 2013

Cyfres gyhoeddiadau

Enw	Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
Cyfrol	7887 LNCS
ISSN (Argraffiad)	0302-9743
ISSN (Electronig)	1611-3349

NDC y CU

Mae’r allbwn hwn yn cyfrannu at y Nod(au) Datblygu Cynaliadwy canlynol

Mynediad at Ddogfen

10.1007/978-3-642-38628-2_29

Cysylltiad parhaol

Dolen barhaus

Ffeiliau eraill a chysylltiadau

Dolen i’r cyhoeddiad yn Scopus

Dyfynnu hyn

Strange, H., Denton, E. R. E., Kibiro, M., & Zwiggelaar, R. (2013). Manifold Learning for Density Segmentation in High Risk Mammograms. Yn J. M. Sanches, L. Micó, & J. S. Cardoso (Gol.), Pattern Recognition and Image Analysis - 6th Iberian Conference, IbPRIA 2013, Proceedings: 6th Iberian Conference, IbPRIA 2013, Funchal, Madeira, Portugal, June 5-7, 2013, Proceedings (tt. 245-252). (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Cyfrol 7887 LNCS). Springer Nature. https://doi.org/10.1007/978-3-642-38628-2_29

Strange, Harry ; Denton, Erika R. E. ; Kibiro, Minnie et al. / Manifold Learning for Density Segmentation in High Risk Mammograms. Pattern Recognition and Image Analysis - 6th Iberian Conference, IbPRIA 2013, Proceedings: 6th Iberian Conference, IbPRIA 2013, Funchal, Madeira, Portugal, June 5-7, 2013, Proceedings. Gol. / João M. Sanches ; Luisa Micó ; Jaime S. Cardoso. Springer Nature, 2013. tt. 245-252 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)).

@inproceedings{d21d24644f414bd3a37e3defd1eb4533,

title = "Manifold Learning for Density Segmentation in High Risk Mammograms",

abstract = "There is a strong correlation between relative mammographic breast density and the risk of developing breast cancer. As such, accurately modelling the percentage of a mammogram that is dense is a pivotal step in density based risk classification. In this work, a novel method based on manifold learning is used to segment high-risk mammograms into density regions. As such, finer details are present in the segmentations and more accurate measures of breast density are produced. A set of high risk (BI-RADS IV) full field digital mammograms with density annotations obtained from radiologists are used to test the validity of the proposed approach. By exploiting the manifold structure of the input space, segmentations with average accuracy of 87% when compared with radiologists{\textquoteright} segmentations can be obtained. This is an increase of over 12% compared with segmentation in the high-dimensional space.",

keywords = "Breast Density Segmentation, Kernel PCA, Mammography, Manifold Learning",

author = "Harry Strange and Denton, {Erika R. E.} and Minnie Kibiro and Reyer Zwiggelaar",

year = "2013",

month = may,

day = "17",

doi = "10.1007/978-3-642-38628-2_29",

language = "English",

isbn = "978-3-642-38627-5",

series = "Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)",

publisher = "Springer Nature",

pages = "245--252",

editor = "Sanches, {Jo{\~a}o M.} and Luisa Mic{\'o} and Cardoso, {Jaime S.}",

booktitle = "Pattern Recognition and Image Analysis - 6th Iberian Conference, IbPRIA 2013, Proceedings",

address = "Switzerland",

}

Strange, H, Denton, ERE, Kibiro, M & Zwiggelaar, R 2013, Manifold Learning for Density Segmentation in High Risk Mammograms. yn JM Sanches, L Micó & JS Cardoso (gol.), Pattern Recognition and Image Analysis - 6th Iberian Conference, IbPRIA 2013, Proceedings: 6th Iberian Conference, IbPRIA 2013, Funchal, Madeira, Portugal, June 5-7, 2013, Proceedings. Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics), cyfrol. 7887 LNCS, Springer Nature, tt. 245-252. https://doi.org/10.1007/978-3-642-38628-2_29

Manifold Learning for Density Segmentation in High Risk Mammograms. / Strange, Harry; Denton, Erika R. E.; Kibiro, Minnie et al.
Pattern Recognition and Image Analysis - 6th Iberian Conference, IbPRIA 2013, Proceedings: 6th Iberian Conference, IbPRIA 2013, Funchal, Madeira, Portugal, June 5-7, 2013, Proceedings. gol. / João M. Sanches; Luisa Micó; Jaime S. Cardoso. Springer Nature, 2013. t. 245-252 (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics); Cyfrol 7887 LNCS).

Allbwn ymchwil: Pennod mewn Llyfr/Adroddiad/Trafodion Cynhadledd › Trafodion Cynhadledd (Nid-Cyfnodolyn fathau)

TY - GEN

T1 - Manifold Learning for Density Segmentation in High Risk Mammograms

AU - Strange, Harry

AU - Denton, Erika R. E.

AU - Kibiro, Minnie

AU - Zwiggelaar, Reyer

PY - 2013/5/17

Y1 - 2013/5/17

N2 - There is a strong correlation between relative mammographic breast density and the risk of developing breast cancer. As such, accurately modelling the percentage of a mammogram that is dense is a pivotal step in density based risk classification. In this work, a novel method based on manifold learning is used to segment high-risk mammograms into density regions. As such, finer details are present in the segmentations and more accurate measures of breast density are produced. A set of high risk (BI-RADS IV) full field digital mammograms with density annotations obtained from radiologists are used to test the validity of the proposed approach. By exploiting the manifold structure of the input space, segmentations with average accuracy of 87% when compared with radiologists’ segmentations can be obtained. This is an increase of over 12% compared with segmentation in the high-dimensional space.

AB - There is a strong correlation between relative mammographic breast density and the risk of developing breast cancer. As such, accurately modelling the percentage of a mammogram that is dense is a pivotal step in density based risk classification. In this work, a novel method based on manifold learning is used to segment high-risk mammograms into density regions. As such, finer details are present in the segmentations and more accurate measures of breast density are produced. A set of high risk (BI-RADS IV) full field digital mammograms with density annotations obtained from radiologists are used to test the validity of the proposed approach. By exploiting the manifold structure of the input space, segmentations with average accuracy of 87% when compared with radiologists’ segmentations can be obtained. This is an increase of over 12% compared with segmentation in the high-dimensional space.

KW - Breast Density Segmentation

KW - Kernel PCA

KW - Mammography

KW - Manifold Learning

UR - http://www.scopus.com/inward/record.url?scp=84883154461&partnerID=8YFLogxK

U2 - 10.1007/978-3-642-38628-2_29

DO - 10.1007/978-3-642-38628-2_29

M3 - Conference Proceeding (Non-Journal item)

SN - 978-3-642-38627-5

T3 - Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)

SP - 245

EP - 252

BT - Pattern Recognition and Image Analysis - 6th Iberian Conference, IbPRIA 2013, Proceedings

A2 - Sanches, João M.

A2 - Micó, Luisa

A2 - Cardoso, Jaime S.

PB - Springer Nature

ER -

Strange H, Denton ERE, Kibiro M, Zwiggelaar R. Manifold Learning for Density Segmentation in High Risk Mammograms. Yn Sanches JM, Micó L, Cardoso JS, golygyddion, Pattern Recognition and Image Analysis - 6th Iberian Conference, IbPRIA 2013, Proceedings: 6th Iberian Conference, IbPRIA 2013, Funchal, Madeira, Portugal, June 5-7, 2013, Proceedings. Springer Nature. 2013. t. 245-252. (Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)). doi: 10.1007/978-3-642-38628-2_29

Manifold Learning for Density Segmentation in High Risk Mammograms

Crynodeb

Cyfres gyhoeddiadau

NDC y CU

Mynediad at Ddogfen

Cysylltiad parhaol

Ffeiliau eraill a chysylltiadau

Ôl bys

Dyfynnu hyn