A Comparison of Regions of Interest in Parenchymal Analysis for Breast Cancer Risk Assessment

Gerson Africano; Otso Arponen; Antti Sassi; Mirja Karivaara-Mäkelä; Kirsi Holli-Helenius; Irina Rinta-Kiikka; Anna-Leena Lääperi; Said Pertuz

doi:10.1109/EMBC44109.2020.9176200

A Comparison of Regions of Interest in Parenchymal Analysis for Breast Cancer Risk Assessment

Gerson Africano, Otso Arponen, Antti Sassi, Mirja Karivaara-Mäkelä, Kirsi Holli-Helenius, Irina Rinta-Kiikka, Anna-Leena Lääperi, Said Pertuz

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › Scientific › peer-review

3 Citations (Scopus)

Abstract

Computerized parenchymal analysis has shown potential to be utilized as an imaging biomarker to estimate the risk of breast cancer. Parenchymal analysis of digital mammograms is based on the extraction of computerized measures to build machine learning-based models for the prediction of breast cancer risk. However, the choice of the region of interest (ROI) for feature extraction within the breast remains an open problem. In this work we perform a comparison between five different methods suggested in the literature for automated ROI selection, including the whole breast (WB), the maximum squared (MS), the retro-areolar region (RA), the lattice-based (LB), and the polar-based (PB) selection methods. For the experiments, we built a retrospective dataset of 896 screening mammograms from 224 women (112 cases and 112 healthy controls). The performance of each ROI selection method was measured in terms of the area under the curve (AUC) values. The AUC values varied between 0.55 and 0.79 depending on the method and experimental settings. The best performance on an independent test set was achieved by the MS method (AUC of 0.59, 95% CI: 0.55-0.64). This method is fully-automated and does not require adjusting hyper-parameters. Based on our results, we prompt the use of the MS method for ROI selection in the computerized parenchymal analysis for breast cancer risk assessment.

Original language	English
Title of host publication	2020 42nd Annual International Conference of the IEEE Engineering in Medicine Biology Society (EMBC)
Publisher	IEEE
Pages	1136-1139
Number of pages	4
ISBN (Electronic)	978-1-7281-1990-8
ISBN (Print)	978-1-7281-1991-5
DOIs	https://doi.org/10.1109/EMBC44109.2020.9176200
Publication status	Published - 2020
Publication type	A4 Article in conference proceedings
Event	Annual International Conference of the IEEE Engineering in Medicine and Biology Society - Montreal, Canada Duration: 20 Jul 2020 → 24 Jul 2020 Conference number: 42nd

Publication series

Name	Annual International Conference of the IEEE Engineering in Medicine and Biology Society
Publisher	IEEE
ISSN (Print)	2375-7477
ISSN (Electronic)	2694-0604

Conference

Conference	Annual International Conference of the IEEE Engineering in Medicine and Biology Society
Country/Territory	Canada
City	Montreal
Period	20/07/20 → 24/07/20

Keywords

cancer
feature extraction
gynaecology
learning (artificial intelligence)
mammography
medical image processing
AUC
area under the curve values
RA
maximum squared
machine learning-based models
computerized measures
digital mammograms
breast cancer risk assessment
computerized parenchymal analysis
MS method
ROI selection method
screening mammograms
polar-based selection methods
lattice-based
retro-areolar region
automated ROI selection
Feature extraction
Breast cancer
Tuning
Risk management
Mammography
Computational modeling
Area Under Curve
Breast Neoplasms
Female
Humans
Retrospective Studies
Risk Assessment

Publication forum classification

Publication forum level 1

UN SDGs

This output contributes to the following UN Sustainable Development Goals (SDGs)

Access to Document

10.1109/EMBC44109.2020.9176200

Cite this

Africano, G., Arponen, O., Sassi, A., Karivaara-Mäkelä, M., Holli-Helenius, K., Rinta-Kiikka, I., Lääperi, A.-L., & Pertuz, S. (2020). A Comparison of Regions of Interest in Parenchymal Analysis for Breast Cancer Risk Assessment. In 2020 42nd Annual International Conference of the IEEE Engineering in Medicine Biology Society (EMBC) (pp. 1136-1139). (Annual International Conference of the IEEE Engineering in Medicine and Biology Society). IEEE. https://doi.org/10.1109/EMBC44109.2020.9176200

Africano, Gerson ; Arponen, Otso ; Sassi, Antti et al. / A Comparison of Regions of Interest in Parenchymal Analysis for Breast Cancer Risk Assessment. 2020 42nd Annual International Conference of the IEEE Engineering in Medicine Biology Society (EMBC). IEEE, 2020. pp. 1136-1139 (Annual International Conference of the IEEE Engineering in Medicine and Biology Society).

@inproceedings{f0ea0c8f5cfe41e8be5813ae7ba20d7a,

title = "A Comparison of Regions of Interest in Parenchymal Analysis for Breast Cancer Risk Assessment",

abstract = "Computerized parenchymal analysis has shown potential to be utilized as an imaging biomarker to estimate the risk of breast cancer. Parenchymal analysis of digital mammograms is based on the extraction of computerized measures to build machine learning-based models for the prediction of breast cancer risk. However, the choice of the region of interest (ROI) for feature extraction within the breast remains an open problem. In this work we perform a comparison between five different methods suggested in the literature for automated ROI selection, including the whole breast (WB), the maximum squared (MS), the retro-areolar region (RA), the lattice-based (LB), and the polar-based (PB) selection methods. For the experiments, we built a retrospective dataset of 896 screening mammograms from 224 women (112 cases and 112 healthy controls). The performance of each ROI selection method was measured in terms of the area under the curve (AUC) values. The AUC values varied between 0.55 and 0.79 depending on the method and experimental settings. The best performance on an independent test set was achieved by the MS method (AUC of 0.59, 95% CI: 0.55-0.64). This method is fully-automated and does not require adjusting hyper-parameters. Based on our results, we prompt the use of the MS method for ROI selection in the computerized parenchymal analysis for breast cancer risk assessment.",

keywords = "cancer, feature extraction, gynaecology, learning (artificial intelligence), mammography, medical image processing, AUC, area under the curve values, RA, maximum squared, machine learning-based models, computerized measures, digital mammograms, breast cancer risk assessment, computerized parenchymal analysis, MS method, ROI selection method, screening mammograms, polar-based selection methods, lattice-based, retro-areolar region, automated ROI selection, Feature extraction, Breast cancer, Tuning, Risk management, Mammography, Computational modeling, Area Under Curve, Breast Neoplasms, Female, Humans, Retrospective Studies, Risk Assessment",

author = "Gerson Africano and Otso Arponen and Antti Sassi and Mirja Karivaara-M{\"a}kel{\"a} and Kirsi Holli-Helenius and Irina Rinta-Kiikka and Anna-Leena L{\"a}{\"a}peri and Said Pertuz",

year = "2020",

doi = "10.1109/EMBC44109.2020.9176200",

language = "English",

isbn = "978-1-7281-1991-5",

series = "Annual International Conference of the IEEE Engineering in Medicine and Biology Society",

publisher = "IEEE",

pages = "1136--1139",

booktitle = "2020 42nd Annual International Conference of the IEEE Engineering in Medicine Biology Society (EMBC)",

address = "United States",

note = "Annual International Conference of the IEEE Engineering in Medicine and Biology Society ; Conference date: 20-07-2020 Through 24-07-2020",

}

Africano, G, Arponen, O, Sassi, A, Karivaara-Mäkelä, M, Holli-Helenius, K , Rinta-Kiikka, I, Lääperi, A-L & Pertuz, S 2020, A Comparison of Regions of Interest in Parenchymal Analysis for Breast Cancer Risk Assessment. in 2020 42nd Annual International Conference of the IEEE Engineering in Medicine Biology Society (EMBC). Annual International Conference of the IEEE Engineering in Medicine and Biology Society, IEEE, pp. 1136-1139, Annual International Conference of the IEEE Engineering in Medicine and Biology Society, Montreal, Canada, 20/07/20. https://doi.org/10.1109/EMBC44109.2020.9176200

A Comparison of Regions of Interest in Parenchymal Analysis for Breast Cancer Risk Assessment. / Africano, Gerson; Arponen, Otso; Sassi, Antti et al.
2020 42nd Annual International Conference of the IEEE Engineering in Medicine Biology Society (EMBC). IEEE, 2020. p. 1136-1139 (Annual International Conference of the IEEE Engineering in Medicine and Biology Society).

Research output: Chapter in Book/Report/Conference proceeding › Conference contribution › Scientific › peer-review

TY - GEN

T1 - A Comparison of Regions of Interest in Parenchymal Analysis for Breast Cancer Risk Assessment

AU - Africano, Gerson

AU - Arponen, Otso

AU - Sassi, Antti

AU - Karivaara-Mäkelä, Mirja

AU - Holli-Helenius, Kirsi

AU - Rinta-Kiikka, Irina

AU - Lääperi, Anna-Leena

AU - Pertuz, Said

N1 - Conference code: 42nd

PY - 2020

Y1 - 2020

N2 - Computerized parenchymal analysis has shown potential to be utilized as an imaging biomarker to estimate the risk of breast cancer. Parenchymal analysis of digital mammograms is based on the extraction of computerized measures to build machine learning-based models for the prediction of breast cancer risk. However, the choice of the region of interest (ROI) for feature extraction within the breast remains an open problem. In this work we perform a comparison between five different methods suggested in the literature for automated ROI selection, including the whole breast (WB), the maximum squared (MS), the retro-areolar region (RA), the lattice-based (LB), and the polar-based (PB) selection methods. For the experiments, we built a retrospective dataset of 896 screening mammograms from 224 women (112 cases and 112 healthy controls). The performance of each ROI selection method was measured in terms of the area under the curve (AUC) values. The AUC values varied between 0.55 and 0.79 depending on the method and experimental settings. The best performance on an independent test set was achieved by the MS method (AUC of 0.59, 95% CI: 0.55-0.64). This method is fully-automated and does not require adjusting hyper-parameters. Based on our results, we prompt the use of the MS method for ROI selection in the computerized parenchymal analysis for breast cancer risk assessment.

AB - Computerized parenchymal analysis has shown potential to be utilized as an imaging biomarker to estimate the risk of breast cancer. Parenchymal analysis of digital mammograms is based on the extraction of computerized measures to build machine learning-based models for the prediction of breast cancer risk. However, the choice of the region of interest (ROI) for feature extraction within the breast remains an open problem. In this work we perform a comparison between five different methods suggested in the literature for automated ROI selection, including the whole breast (WB), the maximum squared (MS), the retro-areolar region (RA), the lattice-based (LB), and the polar-based (PB) selection methods. For the experiments, we built a retrospective dataset of 896 screening mammograms from 224 women (112 cases and 112 healthy controls). The performance of each ROI selection method was measured in terms of the area under the curve (AUC) values. The AUC values varied between 0.55 and 0.79 depending on the method and experimental settings. The best performance on an independent test set was achieved by the MS method (AUC of 0.59, 95% CI: 0.55-0.64). This method is fully-automated and does not require adjusting hyper-parameters. Based on our results, we prompt the use of the MS method for ROI selection in the computerized parenchymal analysis for breast cancer risk assessment.

KW - cancer

KW - feature extraction

KW - gynaecology

KW - learning (artificial intelligence)

KW - mammography

KW - medical image processing

KW - AUC

KW - area under the curve values

KW - RA

KW - maximum squared

KW - machine learning-based models

KW - computerized measures

KW - digital mammograms

KW - breast cancer risk assessment

KW - computerized parenchymal analysis

KW - MS method

KW - ROI selection method

KW - screening mammograms

KW - polar-based selection methods

KW - lattice-based

KW - retro-areolar region

KW - automated ROI selection

KW - Feature extraction

KW - Breast cancer

KW - Tuning

KW - Risk management

KW - Mammography

KW - Computational modeling

KW - Area Under Curve

KW - Breast Neoplasms

KW - Female

KW - Humans

KW - Retrospective Studies

KW - Risk Assessment

U2 - 10.1109/EMBC44109.2020.9176200

DO - 10.1109/EMBC44109.2020.9176200

M3 - Conference contribution

SN - 978-1-7281-1991-5

T3 - Annual International Conference of the IEEE Engineering in Medicine and Biology Society

SP - 1136

EP - 1139

BT - 2020 42nd Annual International Conference of the IEEE Engineering in Medicine Biology Society (EMBC)

PB - IEEE

T2 - Annual International Conference of the IEEE Engineering in Medicine and Biology Society

Y2 - 20 July 2020 through 24 July 2020

ER -

Africano G, Arponen O, Sassi A, Karivaara-Mäkelä M, Holli-Helenius K , Rinta-Kiikka I et al. A Comparison of Regions of Interest in Parenchymal Analysis for Breast Cancer Risk Assessment. In 2020 42nd Annual International Conference of the IEEE Engineering in Medicine Biology Society (EMBC). IEEE. 2020. p. 1136-1139. (Annual International Conference of the IEEE Engineering in Medicine and Biology Society). doi: 10.1109/EMBC44109.2020.9176200

A Comparison of Regions of Interest in Parenchymal Analysis for Breast Cancer Risk Assessment

Abstract

Publication series

Conference

Keywords

Publication forum classification

UN SDGs

Access to Document

Fingerprint

Cite this