Using deep learning for mammography classification

Pinar Uskaner Hepsaǧ, Selma Ayşe Özel, Adnan Yazici

Research output: Chapter in Book/Report/Conference proceedingConference contribution

5 Citations (Scopus)

Abstract

Breast biopsies based on the results of mammography and ultrasound have been diagnosed as benign at a rate of approximately 40 to 60 percent. Negative biopsy results have negative impacts on many aspects such as unnecessary operations, fear, pain, and cost. Therefore, there is a need for a more reliable technique to reduce the number of unnecessary biopsies in the diagnosis of breast cancer. So, computer-aided diagnostic methods are very important for doctors to make more accurate decisions and to avoid unnecessary biopsies. For this purpose, we apply deep learning using Convolutional Neural Networks (CNN) to classify abnormalities as benign or malignant in mammogram images by using two different databases namely, mini-MIAS and BCDR. While mini-MIAS database has valuable information like location of the center of abnormality and radius of the circle that surrounds the abnormality, BCDR database does not have. When we use both dataset as they are, we observe accuracy, precision, recall, and f-score values between around 60% and 72%. In order to improve our results, we take the benefit of preprocessing methods containing cropping, augmentation, and balancing image data. In an effort to crop image data sourced from BCDR, we create a mask to find region of interest. After applying our preprocessing methods over the BCDR dataset, we observe that classification accuracy improves from 65% to around 85%. When we compare the classification accuracy, precision, recall and f-score obtained from the MIAS database with those obtained from the BCDR database we found that after applying preprocessing methods to BCDR dataset, the classification performance become very close to each other for the two datasets.

Original languageEnglish
Title of host publication2nd International Conference on Computer Science and Engineering, UBMK 2017
PublisherInstitute of Electrical and Electronics Engineers Inc.
Pages418-423
Number of pages6
ISBN (Electronic)9781538609309
DOIs
Publication statusPublished - Oct 31 2017
Event2nd International Conference on Computer Science and Engineering, UBMK 2017 - Antalya, Turkey
Duration: Oct 5 2017Oct 8 2017

Publication series

Name2nd International Conference on Computer Science and Engineering, UBMK 2017

Conference

Conference2nd International Conference on Computer Science and Engineering, UBMK 2017
CountryTurkey
CityAntalya
Period10/5/1710/8/17

Fingerprint

Mammography
Biopsy
Crops
Masks
Ultrasonics
Deep learning
Neural networks
Costs

Keywords

  • BCDR
  • Breast Cancer
  • Convolutional Neural Networks
  • Deep Learning
  • Mammogram Classification
  • MIAS

ASJC Scopus subject areas

  • Artificial Intelligence
  • Computer Networks and Communications
  • Computer Science Applications
  • Computer Vision and Pattern Recognition

Cite this

Hepsaǧ, P. U., Özel, S. A., & Yazici, A. (2017). Using deep learning for mammography classification. In 2nd International Conference on Computer Science and Engineering, UBMK 2017 (pp. 418-423). [8093429] (2nd International Conference on Computer Science and Engineering, UBMK 2017). Institute of Electrical and Electronics Engineers Inc.. https://doi.org/10.1109/UBMK.2017.8093429

Using deep learning for mammography classification. / Hepsaǧ, Pinar Uskaner; Özel, Selma Ayşe; Yazici, Adnan.

2nd International Conference on Computer Science and Engineering, UBMK 2017. Institute of Electrical and Electronics Engineers Inc., 2017. p. 418-423 8093429 (2nd International Conference on Computer Science and Engineering, UBMK 2017).

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Hepsaǧ, PU, Özel, SA & Yazici, A 2017, Using deep learning for mammography classification. in 2nd International Conference on Computer Science and Engineering, UBMK 2017., 8093429, 2nd International Conference on Computer Science and Engineering, UBMK 2017, Institute of Electrical and Electronics Engineers Inc., pp. 418-423, 2nd International Conference on Computer Science and Engineering, UBMK 2017, Antalya, Turkey, 10/5/17. https://doi.org/10.1109/UBMK.2017.8093429
Hepsaǧ PU, Özel SA, Yazici A. Using deep learning for mammography classification. In 2nd International Conference on Computer Science and Engineering, UBMK 2017. Institute of Electrical and Electronics Engineers Inc. 2017. p. 418-423. 8093429. (2nd International Conference on Computer Science and Engineering, UBMK 2017). https://doi.org/10.1109/UBMK.2017.8093429
Hepsaǧ, Pinar Uskaner ; Özel, Selma Ayşe ; Yazici, Adnan. / Using deep learning for mammography classification. 2nd International Conference on Computer Science and Engineering, UBMK 2017. Institute of Electrical and Electronics Engineers Inc., 2017. pp. 418-423 (2nd International Conference on Computer Science and Engineering, UBMK 2017).
@inproceedings{230df0e6731c48bb8af9af73632443fa,
title = "Using deep learning for mammography classification",
abstract = "Breast biopsies based on the results of mammography and ultrasound have been diagnosed as benign at a rate of approximately 40 to 60 percent. Negative biopsy results have negative impacts on many aspects such as unnecessary operations, fear, pain, and cost. Therefore, there is a need for a more reliable technique to reduce the number of unnecessary biopsies in the diagnosis of breast cancer. So, computer-aided diagnostic methods are very important for doctors to make more accurate decisions and to avoid unnecessary biopsies. For this purpose, we apply deep learning using Convolutional Neural Networks (CNN) to classify abnormalities as benign or malignant in mammogram images by using two different databases namely, mini-MIAS and BCDR. While mini-MIAS database has valuable information like location of the center of abnormality and radius of the circle that surrounds the abnormality, BCDR database does not have. When we use both dataset as they are, we observe accuracy, precision, recall, and f-score values between around 60{\%} and 72{\%}. In order to improve our results, we take the benefit of preprocessing methods containing cropping, augmentation, and balancing image data. In an effort to crop image data sourced from BCDR, we create a mask to find region of interest. After applying our preprocessing methods over the BCDR dataset, we observe that classification accuracy improves from 65{\%} to around 85{\%}. When we compare the classification accuracy, precision, recall and f-score obtained from the MIAS database with those obtained from the BCDR database we found that after applying preprocessing methods to BCDR dataset, the classification performance become very close to each other for the two datasets.",
keywords = "BCDR, Breast Cancer, Convolutional Neural Networks, Deep Learning, Mammogram Classification, MIAS",
author = "Hepsaǧ, {Pinar Uskaner} and {\"O}zel, {Selma Ayşe} and Adnan Yazici",
year = "2017",
month = "10",
day = "31",
doi = "10.1109/UBMK.2017.8093429",
language = "English",
series = "2nd International Conference on Computer Science and Engineering, UBMK 2017",
publisher = "Institute of Electrical and Electronics Engineers Inc.",
pages = "418--423",
booktitle = "2nd International Conference on Computer Science and Engineering, UBMK 2017",
address = "United States",

}

TY - GEN

T1 - Using deep learning for mammography classification

AU - Hepsaǧ, Pinar Uskaner

AU - Özel, Selma Ayşe

AU - Yazici, Adnan

PY - 2017/10/31

Y1 - 2017/10/31

N2 - Breast biopsies based on the results of mammography and ultrasound have been diagnosed as benign at a rate of approximately 40 to 60 percent. Negative biopsy results have negative impacts on many aspects such as unnecessary operations, fear, pain, and cost. Therefore, there is a need for a more reliable technique to reduce the number of unnecessary biopsies in the diagnosis of breast cancer. So, computer-aided diagnostic methods are very important for doctors to make more accurate decisions and to avoid unnecessary biopsies. For this purpose, we apply deep learning using Convolutional Neural Networks (CNN) to classify abnormalities as benign or malignant in mammogram images by using two different databases namely, mini-MIAS and BCDR. While mini-MIAS database has valuable information like location of the center of abnormality and radius of the circle that surrounds the abnormality, BCDR database does not have. When we use both dataset as they are, we observe accuracy, precision, recall, and f-score values between around 60% and 72%. In order to improve our results, we take the benefit of preprocessing methods containing cropping, augmentation, and balancing image data. In an effort to crop image data sourced from BCDR, we create a mask to find region of interest. After applying our preprocessing methods over the BCDR dataset, we observe that classification accuracy improves from 65% to around 85%. When we compare the classification accuracy, precision, recall and f-score obtained from the MIAS database with those obtained from the BCDR database we found that after applying preprocessing methods to BCDR dataset, the classification performance become very close to each other for the two datasets.

AB - Breast biopsies based on the results of mammography and ultrasound have been diagnosed as benign at a rate of approximately 40 to 60 percent. Negative biopsy results have negative impacts on many aspects such as unnecessary operations, fear, pain, and cost. Therefore, there is a need for a more reliable technique to reduce the number of unnecessary biopsies in the diagnosis of breast cancer. So, computer-aided diagnostic methods are very important for doctors to make more accurate decisions and to avoid unnecessary biopsies. For this purpose, we apply deep learning using Convolutional Neural Networks (CNN) to classify abnormalities as benign or malignant in mammogram images by using two different databases namely, mini-MIAS and BCDR. While mini-MIAS database has valuable information like location of the center of abnormality and radius of the circle that surrounds the abnormality, BCDR database does not have. When we use both dataset as they are, we observe accuracy, precision, recall, and f-score values between around 60% and 72%. In order to improve our results, we take the benefit of preprocessing methods containing cropping, augmentation, and balancing image data. In an effort to crop image data sourced from BCDR, we create a mask to find region of interest. After applying our preprocessing methods over the BCDR dataset, we observe that classification accuracy improves from 65% to around 85%. When we compare the classification accuracy, precision, recall and f-score obtained from the MIAS database with those obtained from the BCDR database we found that after applying preprocessing methods to BCDR dataset, the classification performance become very close to each other for the two datasets.

KW - BCDR

KW - Breast Cancer

KW - Convolutional Neural Networks

KW - Deep Learning

KW - Mammogram Classification

KW - MIAS

UR - http://www.scopus.com/inward/record.url?scp=85040535746&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85040535746&partnerID=8YFLogxK

U2 - 10.1109/UBMK.2017.8093429

DO - 10.1109/UBMK.2017.8093429

M3 - Conference contribution

AN - SCOPUS:85040535746

T3 - 2nd International Conference on Computer Science and Engineering, UBMK 2017

SP - 418

EP - 423

BT - 2nd International Conference on Computer Science and Engineering, UBMK 2017

PB - Institute of Electrical and Electronics Engineers Inc.

ER -