Condensed U-Net (CU-Net): An Improved U-Net Architecture for
Cell Segmentation Powered by 4x4 Max-Pooling Layers

D 2020

Condensed U-Net (CU-Net): An Improved U-Net Architecture for Cell Segmentation Powered by 4x4 Max-Pooling Layers

AKBAS, Cem Emre and Michal KOZUBEK

Basic information

Original name

Condensed U-Net (CU-Net): An Improved U-Net Architecture for Cell Segmentation Powered by 4x4 Max-Pooling Layers

Authors

AKBAS, Cem Emre (792 Turkey, belonging to the institution) and Michal KOZUBEK (203 Czech Republic, guarantor, belonging to the institution)

Edition

Iowa, IEEE 17th International Symposium on Biomedical Imaging, p. 446-450, 5 pp. 2020

Publisher

IEEE

Other information

Language

English

Type of outcome

Stať ve sborníku

Field of Study

10201 Computer sciences, information science, bioinformatics

Country of publisher

United States of America

Confidentiality degree

není předmětem státního či obchodního tajemství

Publication form

electronic version available online

References:

URL

RIV identification code

RIV/00216224:14330/20:00115503

Organization unit

Faculty of Informatics

ISBN

978-1-5386-9330-8

ISSN

DOI

http://dx.doi.org/10.1109/ISBI45749.2020.9098351

UT WoS

000578080300083

Keywords in English

Biomedical Image Segmentation; Convolutional Neural Networks; Deep Learning; Feature Learning; Max Pooling

Abstract

V originále

Recently, the U-Net has been the dominant approach in the cell segmentation task in biomedical images due to its success in a wide range of image recognition tasks. However, recent studies did not focus enough on updating the architecture of the U-Net and designing specialized loss functions for bioimage segmentation. We show that the U-Net architecture can achieve more successful results with efficient architectural improvements. We propose a condensed encoder-decoder scheme that employs the 4x4 max-pooling operation and triple convolutional layers. The proposed network architecture is trained using a novel combined loss function specifically designed for bioimage segmentation. On the benchmark datasets from the Cell Tracking Challenge, the experimental results show that the proposed cell segmentation system outperforms the U-Net.

Links

EF16_013/0001775, research and development project

Name: Modernizace a podpora výzkumných aktivit národní infrastruktury pro biologické a medicínské zobrazování Czech-BioImaging

LTC17016, research and development project

Name: Benchmarking algoritmů segmentace a sledování buněk

Investor: Ministry of Education, Youth and Sports of the CR, Benchmarking of algorithms for cell segmentation and tracking, INTER-COST

MUNI/A/1050/2019, interní kód MU

Name: Rozsáhlé výpočetní systémy: modely, aplikace a verifikace IX (Acronym: SV-FI MAV IX)

Investor: Masaryk University, Category A

Citovat

AKBAS, Cem Emre and Michal KOZUBEK. Condensed U-Net (CU-Net): An Improved U-Net Architecture for Cell Segmentation Powered by 4x4 Max-Pooling Layers. Online. In IEEE 17th International Symposium on Biomedical Imaging. Iowa: IEEE, 2020, p. 446-450. ISBN 978-1-5386-9330-8. Available from: https://dx.doi.org/10.1109/ISBI45749.2020.9098351.

@inproceedings{1644098,
   author = {Akbas, Cem Emre and Kozubek, Michal},
   address = {Iowa},
   booktitle = {IEEE 17th International Symposium on Biomedical Imaging},
   doi = {http://dx.doi.org/10.1109/ISBI45749.2020.9098351},
   keywords = {Biomedical Image Segmentation; Convolutional Neural Networks; Deep Learning; Feature Learning; Max Pooling},
   howpublished = {elektronická verze "online"},
   language = {eng},
   location = {Iowa},
   isbn = {978-1-5386-9330-8},
   pages = {446-450},
   publisher = {IEEE},
   title = {Condensed U-Net (CU-Net): An Improved U-Net Architecture for Cell Segmentation Powered by 4x4 Max-Pooling Layers},
   url = {https://ieeexplore.ieee.org/abstract/document/9098351},
   year = {2020}
}

TY  - JOUR
ID  - 1644098
AU  - Akbas, Cem Emre - Kozubek, Michal
PY  - 2020
TI  - Condensed U-Net (CU-Net): An Improved U-Net Architecture for Cell Segmentation Powered by 4x4 Max-Pooling Layers
PB  - IEEE
CY  - Iowa
SN  - 9781538693308
KW  - Biomedical Image Segmentation
KW  - Convolutional Neural Networks
KW  - Deep Learning
KW  - Feature Learning
KW  - Max Pooling
UR  - https://ieeexplore.ieee.org/abstract/document/9098351
L2  - https://ieeexplore.ieee.org/abstract/document/9098351
N2  - Recently, the U-Net has been the dominant approach in the cell segmentation task in biomedical images due to its success in a wide range of image recognition tasks. However, recent studies did not focus enough on updating the architecture of the U-Net and designing specialized loss functions for bioimage segmentation. We show that the U-Net architecture can achieve more successful results with efficient architectural improvements. We propose a condensed encoder-decoder scheme that employs the 4x4 max-pooling operation and triple convolutional layers. The proposed network architecture is trained using a novel combined loss function specifically designed for bioimage segmentation. On the benchmark datasets from the Cell Tracking Challenge, the experimental results show that the proposed cell segmentation system outperforms the U-Net.
ER  -

AKBAS, Cem Emre and Michal KOZUBEK. Condensed U-Net (CU-Net): An Improved U-Net Architecture for Cell Segmentation Powered by 4x4 Max-Pooling Layers. Online. In \textit{IEEE 17th International Symposium on Biomedical Imaging}. Iowa: IEEE, 2020, p.~446-450. ISBN~978-1-5386-9330-8. Available from: https://dx.doi.org/10.1109/ISBI45749.2020.9098351.

Detailed Information on Publication Record