Automated and real-time segmentation of suspicious breast masses using convolutional neural network

Viksit Kumar, Jeremy M. Webb, Adriana Gregory, Max Denis, Duane D. Meixner, Mahdi Bayat, Dana H. Whaley, Mostafa Fatemi, Azra Alizad

Research output: Contribution to journalArticle

4 Citations (Scopus)

Abstract

In this work, a computer-aided tool for detection was developed to segment breast masses from clinical ultrasound (US) scans. The underlying Multi U-net algorithm is based on convolutional neural networks. Under the Mayo Clinic Institutional Review Board protocol, a prospective study of the automatic segmentation of suspicious breast masses was performed. The cohort consisted of 258 female patients who were clinically identified with suspicious breast masses and underwent clinical US scan and breast biopsy. The computer-aided detection tool effectively segmented the breast masses, achieving a mean Dice coefficient of 0.82, a true positive fraction (TPF) of 0.84, and a false positive fraction (FPF) of 0.01. By avoiding positioning of an initial seed, the algorithm is able to segment images in real time (13–55 ms per image), and can have potential clinical applications. The algorithm is at par with a conventional seeded algorithm, which had a mean Dice coefficient of 0.84 and performs significantly better (P< 0.0001) than the original U-net algorithm.

Original languageEnglish (US)
Article numbere0195816
JournalPLoS One
Volume13
Issue number5
DOIs
StatePublished - May 1 2018

Fingerprint

neural networks
breasts
Breast
Neural networks
Ultrasonics
Biopsy
Research Ethics Committees
prospective studies
biopsy
Prospective Studies
seeds

ASJC Scopus subject areas

  • Biochemistry, Genetics and Molecular Biology(all)
  • Agricultural and Biological Sciences(all)

Cite this

Automated and real-time segmentation of suspicious breast masses using convolutional neural network. / Kumar, Viksit; Webb, Jeremy M.; Gregory, Adriana; Denis, Max; Meixner, Duane D.; Bayat, Mahdi; Whaley, Dana H.; Fatemi, Mostafa; Alizad, Azra.

In: PLoS One, Vol. 13, No. 5, e0195816, 01.05.2018.

Research output: Contribution to journalArticle

Kumar, Viksit ; Webb, Jeremy M. ; Gregory, Adriana ; Denis, Max ; Meixner, Duane D. ; Bayat, Mahdi ; Whaley, Dana H. ; Fatemi, Mostafa ; Alizad, Azra. / Automated and real-time segmentation of suspicious breast masses using convolutional neural network. In: PLoS One. 2018 ; Vol. 13, No. 5.
@article{5965f08da4c847c18ff2c609bc7c2e62,
title = "Automated and real-time segmentation of suspicious breast masses using convolutional neural network",
abstract = "In this work, a computer-aided tool for detection was developed to segment breast masses from clinical ultrasound (US) scans. The underlying Multi U-net algorithm is based on convolutional neural networks. Under the Mayo Clinic Institutional Review Board protocol, a prospective study of the automatic segmentation of suspicious breast masses was performed. The cohort consisted of 258 female patients who were clinically identified with suspicious breast masses and underwent clinical US scan and breast biopsy. The computer-aided detection tool effectively segmented the breast masses, achieving a mean Dice coefficient of 0.82, a true positive fraction (TPF) of 0.84, and a false positive fraction (FPF) of 0.01. By avoiding positioning of an initial seed, the algorithm is able to segment images in real time (13–55 ms per image), and can have potential clinical applications. The algorithm is at par with a conventional seeded algorithm, which had a mean Dice coefficient of 0.84 and performs significantly better (P< 0.0001) than the original U-net algorithm.",
author = "Viksit Kumar and Webb, {Jeremy M.} and Adriana Gregory and Max Denis and Meixner, {Duane D.} and Mahdi Bayat and Whaley, {Dana H.} and Mostafa Fatemi and Azra Alizad",
year = "2018",
month = "5",
day = "1",
doi = "10.1371/journal.pone.0195816",
language = "English (US)",
volume = "13",
journal = "PLoS One",
issn = "1932-6203",
publisher = "Public Library of Science",
number = "5",

}

TY - JOUR

T1 - Automated and real-time segmentation of suspicious breast masses using convolutional neural network

AU - Kumar, Viksit

AU - Webb, Jeremy M.

AU - Gregory, Adriana

AU - Denis, Max

AU - Meixner, Duane D.

AU - Bayat, Mahdi

AU - Whaley, Dana H.

AU - Fatemi, Mostafa

AU - Alizad, Azra

PY - 2018/5/1

Y1 - 2018/5/1

N2 - In this work, a computer-aided tool for detection was developed to segment breast masses from clinical ultrasound (US) scans. The underlying Multi U-net algorithm is based on convolutional neural networks. Under the Mayo Clinic Institutional Review Board protocol, a prospective study of the automatic segmentation of suspicious breast masses was performed. The cohort consisted of 258 female patients who were clinically identified with suspicious breast masses and underwent clinical US scan and breast biopsy. The computer-aided detection tool effectively segmented the breast masses, achieving a mean Dice coefficient of 0.82, a true positive fraction (TPF) of 0.84, and a false positive fraction (FPF) of 0.01. By avoiding positioning of an initial seed, the algorithm is able to segment images in real time (13–55 ms per image), and can have potential clinical applications. The algorithm is at par with a conventional seeded algorithm, which had a mean Dice coefficient of 0.84 and performs significantly better (P< 0.0001) than the original U-net algorithm.

AB - In this work, a computer-aided tool for detection was developed to segment breast masses from clinical ultrasound (US) scans. The underlying Multi U-net algorithm is based on convolutional neural networks. Under the Mayo Clinic Institutional Review Board protocol, a prospective study of the automatic segmentation of suspicious breast masses was performed. The cohort consisted of 258 female patients who were clinically identified with suspicious breast masses and underwent clinical US scan and breast biopsy. The computer-aided detection tool effectively segmented the breast masses, achieving a mean Dice coefficient of 0.82, a true positive fraction (TPF) of 0.84, and a false positive fraction (FPF) of 0.01. By avoiding positioning of an initial seed, the algorithm is able to segment images in real time (13–55 ms per image), and can have potential clinical applications. The algorithm is at par with a conventional seeded algorithm, which had a mean Dice coefficient of 0.84 and performs significantly better (P< 0.0001) than the original U-net algorithm.

UR - http://www.scopus.com/inward/record.url?scp=85047194162&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=85047194162&partnerID=8YFLogxK

U2 - 10.1371/journal.pone.0195816

DO - 10.1371/journal.pone.0195816

M3 - Article

VL - 13

JO - PLoS One

JF - PLoS One

SN - 1932-6203

IS - 5

M1 - e0195816

ER -