SD-CNN

A shallow-deep CNN for improved breast cancer diagnosis

Fei Gao, Teresa Wu, Jing Li, Bin Zheng, Lingxiang Ruan, Desheng Shang, Bhavika Patel

Research output: Contribution to journalArticle

2 Citations (Scopus)

Abstract

Breast cancer is the second leading cause of cancer death among women worldwide. Nevertheless, it is also one of the most treatable malignances if detected early. Screening for breast cancer with full field digital mammography (FFDM) has been widely used. However, it demonstrates limited performance for women with dense breasts. An emerging technology in the field is contrast-enhanced digital mammography (CEDM), which includes a low energy (LE) image similar to FFDM, and a recombined image leveraging tumor neoangiogenesis similar to breast magnetic resonance imaging (MRI). CEDM has shown better diagnostic accuracy than FFDM. While promising, CEDM is not yet widely available across medical centers. In this research, we propose a Shallow-Deep Convolutional Neural Network (SD-CNN) where a shallow CNN is developed to derive “virtual” recombined images from LE images, and a deep CNN is employed to extract novel features from LE, recombined or “virtual” recombined images for ensemble models to classify the cases as benign vs. cancer. To evaluate the validity of our approach, we first develop a deep-CNN using 49 CEDM cases collected from Mayo Clinic to prove the contributions from recombined images for improved breast cancer diagnosis (0.85 in accuracy, 0.84 in AUC using LE imaging vs. 0.89 in accuracy, 0.91 in AUC using both LE and recombined imaging). We then develop a shallow-CNN using the same 49 CEDM cases to learn the nonlinear mapping from LE to recombined images. Next, we use 89 FFDM cases from INbreast, a public database to generate “virtual” recombined images. Using FFDM alone provides 0.84 in accuracy (AUC = 0.87), whereas SD-CNN improves the diagnostic accuracy to 0.90 (AUC = 0.92).

Original languageEnglish (US)
Pages (from-to)53-62
Number of pages10
JournalComputerized Medical Imaging and Graphics
Volume70
DOIs
StatePublished - Dec 1 2018

Fingerprint

Mammography
Breast Neoplasms
Neural networks
Area Under Curve
Imaging techniques
Breast
Neoplasms
Magnetic resonance
Tumors
Cause of Death
Screening
Magnetic Resonance Imaging
Databases
Technology

Keywords

  • Breast tumor
  • CEDM
  • Classification
  • Deep learning
  • Digital mammography
  • Image synthesis

ASJC Scopus subject areas

  • Radiological and Ultrasound Technology
  • Radiology Nuclear Medicine and imaging
  • Computer Vision and Pattern Recognition
  • Health Informatics
  • Computer Graphics and Computer-Aided Design

Cite this

SD-CNN : A shallow-deep CNN for improved breast cancer diagnosis. / Gao, Fei; Wu, Teresa; Li, Jing; Zheng, Bin; Ruan, Lingxiang; Shang, Desheng; Patel, Bhavika.

In: Computerized Medical Imaging and Graphics, Vol. 70, 01.12.2018, p. 53-62.

Research output: Contribution to journalArticle

Gao, Fei ; Wu, Teresa ; Li, Jing ; Zheng, Bin ; Ruan, Lingxiang ; Shang, Desheng ; Patel, Bhavika. / SD-CNN : A shallow-deep CNN for improved breast cancer diagnosis. In: Computerized Medical Imaging and Graphics. 2018 ; Vol. 70. pp. 53-62.
@article{497df0476b634bdd9bf0523ea3101b99,
title = "SD-CNN: A shallow-deep CNN for improved breast cancer diagnosis",
abstract = "Breast cancer is the second leading cause of cancer death among women worldwide. Nevertheless, it is also one of the most treatable malignances if detected early. Screening for breast cancer with full field digital mammography (FFDM) has been widely used. However, it demonstrates limited performance for women with dense breasts. An emerging technology in the field is contrast-enhanced digital mammography (CEDM), which includes a low energy (LE) image similar to FFDM, and a recombined image leveraging tumor neoangiogenesis similar to breast magnetic resonance imaging (MRI). CEDM has shown better diagnostic accuracy than FFDM. While promising, CEDM is not yet widely available across medical centers. In this research, we propose a Shallow-Deep Convolutional Neural Network (SD-CNN) where a shallow CNN is developed to derive “virtual” recombined images from LE images, and a deep CNN is employed to extract novel features from LE, recombined or “virtual” recombined images for ensemble models to classify the cases as benign vs. cancer. To evaluate the validity of our approach, we first develop a deep-CNN using 49 CEDM cases collected from Mayo Clinic to prove the contributions from recombined images for improved breast cancer diagnosis (0.85 in accuracy, 0.84 in AUC using LE imaging vs. 0.89 in accuracy, 0.91 in AUC using both LE and recombined imaging). We then develop a shallow-CNN using the same 49 CEDM cases to learn the nonlinear mapping from LE to recombined images. Next, we use 89 FFDM cases from INbreast, a public database to generate “virtual” recombined images. Using FFDM alone provides 0.84 in accuracy (AUC = 0.87), whereas SD-CNN improves the diagnostic accuracy to 0.90 (AUC = 0.92).",
keywords = "Breast tumor, CEDM, Classification, Deep learning, Digital mammography, Image synthesis",
author = "Fei Gao and Teresa Wu and Jing Li and Bin Zheng and Lingxiang Ruan and Desheng Shang and Bhavika Patel",
year = "2018",
month = "12",
day = "1",
doi = "10.1016/j.compmedimag.2018.09.004",
language = "English (US)",
volume = "70",
pages = "53--62",
journal = "Computerized Medical Imaging and Graphics",
issn = "0895-6111",
publisher = "Elsevier Limited",

}

TY - JOUR

T1 - SD-CNN

T2 - A shallow-deep CNN for improved breast cancer diagnosis

AU - Gao, Fei

AU - Wu, Teresa

AU - Li, Jing

AU - Zheng, Bin

AU - Ruan, Lingxiang

AU - Shang, Desheng

AU - Patel, Bhavika

PY - 2018/12/1

Y1 - 2018/12/1

N2 - Breast cancer is the second leading cause of cancer death among women worldwide. Nevertheless, it is also one of the most treatable malignances if detected early. Screening for breast cancer with full field digital mammography (FFDM) has been widely used. However, it demonstrates limited performance for women with dense breasts. An emerging technology in the field is contrast-enhanced digital mammography (CEDM), which includes a low energy (LE) image similar to FFDM, and a recombined image leveraging tumor neoangiogenesis similar to breast magnetic resonance imaging (MRI). CEDM has shown better diagnostic accuracy than FFDM. While promising, CEDM is not yet widely available across medical centers. In this research, we propose a Shallow-Deep Convolutional Neural Network (SD-CNN) where a shallow CNN is developed to derive “virtual” recombined images from LE images, and a deep CNN is employed to extract novel features from LE, recombined or “virtual” recombined images for ensemble models to classify the cases as benign vs. cancer. To evaluate the validity of our approach, we first develop a deep-CNN using 49 CEDM cases collected from Mayo Clinic to prove the contributions from recombined images for improved breast cancer diagnosis (0.85 in accuracy, 0.84 in AUC using LE imaging vs. 0.89 in accuracy, 0.91 in AUC using both LE and recombined imaging). We then develop a shallow-CNN using the same 49 CEDM cases to learn the nonlinear mapping from LE to recombined images. Next, we use 89 FFDM cases from INbreast, a public database to generate “virtual” recombined images. Using FFDM alone provides 0.84 in accuracy (AUC = 0.87), whereas SD-CNN improves the diagnostic accuracy to 0.90 (AUC = 0.92).

AB - Breast cancer is the second leading cause of cancer death among women worldwide. Nevertheless, it is also one of the most treatable malignances if detected early. Screening for breast cancer with full field digital mammography (FFDM) has been widely used. However, it demonstrates limited performance for women with dense breasts. An emerging technology in the field is contrast-enhanced digital mammography (CEDM), which includes a low energy (LE) image similar to FFDM, and a recombined image leveraging tumor neoangiogenesis similar to breast magnetic resonance imaging (MRI). CEDM has shown better diagnostic accuracy than FFDM. While promising, CEDM is not yet widely available across medical centers. In this research, we propose a Shallow-Deep Convolutional Neural Network (SD-CNN) where a shallow CNN is developed to derive “virtual” recombined images from LE images, and a deep CNN is employed to extract novel features from LE, recombined or “virtual” recombined images for ensemble models to classify the cases as benign vs. cancer. To evaluate the validity of our approach, we first develop a deep-CNN using 49 CEDM cases collected from Mayo Clinic to prove the contributions from recombined images for improved breast cancer diagnosis (0.85 in accuracy, 0.84 in AUC using LE imaging vs. 0.89 in accuracy, 0.91 in AUC using both LE and recombined imaging). We then develop a shallow-CNN using the same 49 CEDM cases to learn the nonlinear mapping from LE to recombined images. Next, we use 89 FFDM cases from INbreast, a public database to generate “virtual” recombined images. Using FFDM alone provides 0.84 in accuracy (AUC = 0.87), whereas SD-CNN improves the diagnostic accuracy to 0.90 (AUC = 0.92).

KW - Breast tumor

KW - CEDM

KW - Classification

KW - Deep learning

KW - Digital mammography

KW - Image synthesis

UR - http://www.scopus.com/inward/record.url?scp=85054177420&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85054177420&partnerID=8YFLogxK

U2 - 10.1016/j.compmedimag.2018.09.004

DO - 10.1016/j.compmedimag.2018.09.004

M3 - Article

VL - 70

SP - 53

EP - 62

JO - Computerized Medical Imaging and Graphics

JF - Computerized Medical Imaging and Graphics

SN - 0895-6111

ER -