Mask-RCNN and u-net ensembled for nuclei segmentation

Aarno Oskar Vuola; Saad Ullah Akram; Juho Kannala

doi:10.1109/ISBI.2019.8759574

Mask-RCNN and u-net ensembled for nuclei segmentation

Aarno Oskar Vuola^*
, Saad Ullah Akram
, Juho Kannala

^*Corresponding author for this work

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

211 Citations (Scopus)

Abstract

Nuclei segmentation is both an important and in some ways ideal task for modern computer vision methods, e.g. convolutional neural networks. While recent developments in theory and open-source software have made these tools easier to implement, expert knowledge is still required to choose the Sight model architecture and training setup. We compare two popular segmentation frameworks, U-Net and Mask-RCNN in the nuclei segmentation task and find that they have different strengths and failures. To get the best of both worlds, we develop an ensemble model to combine their predictions that can outperform both models by a significant margin and should be considered when aiming for best nuclei segmentation performance.

Original language	English
Title of host publication	ISBI 2019 - 2019 IEEE International Symposium on Biomedical Imaging
Publisher	IEEE
Pages	208-212
Number of pages	5
ISBN (Electronic)	9781538636411
DOIs	https://doi.org/10.1109/ISBI.2019.8759574
Publication status	Published - 2019
MoE publication type	A4 Conference publication
Event	IEEE International Symposium on Biomedical Imaging - Venice, Italy Duration: 8 Apr 2019 → 11 Apr 2019 Conference number: 16

Publication series

Name	IEEE International Symposium on Biomedical Imaging
Publisher	IEEE
ISSN (Print)	1945-7928

Conference

Conference	IEEE International Symposium on Biomedical Imaging
Abbreviated title	ISBI
Country/Territory	Italy
City	Venice
Period	08/04/2019 → 11/04/2019

Keywords

nuclei segmentation
microscopy image analysis
convolutional neural networks

Access to Document

10.1109/ISBI.2019.8759574

https://arxiv.org/abs/1901.10170

Cite this

@inproceedings{83453e54536e40c6bbf0090b6f3d2967,

title = "Mask-RCNN and u-net ensembled for nuclei segmentation",

abstract = "Nuclei segmentation is both an important and in some ways ideal task for modern computer vision methods, e.g. convolutional neural networks. While recent developments in theory and open-source software have made these tools easier to implement, expert knowledge is still required to choose the Sight model architecture and training setup. We compare two popular segmentation frameworks, U-Net and Mask-RCNN in the nuclei segmentation task and find that they have different strengths and failures. To get the best of both worlds, we develop an ensemble model to combine their predictions that can outperform both models by a significant margin and should be considered when aiming for best nuclei segmentation performance.",

keywords = "nuclei segmentation, microscopy image analysis, convolutional neural networks",

author = "Vuola, \{Aarno Oskar\} and Akram, \{Saad Ullah\} and Juho Kannala",

year = "2019",

doi = "10.1109/ISBI.2019.8759574",

language = "English",

series = "IEEE International Symposium on Biomedical Imaging",

publisher = "IEEE",

pages = "208--212",

booktitle = "ISBI 2019 - 2019 IEEE International Symposium on Biomedical Imaging",

address = "United States",

note = "IEEE International Symposium on Biomedical Imaging, ISBI ; Conference date: 08-04-2019 Through 11-04-2019",

}

Mask-RCNN and u-net ensembled for nuclei segmentation. / Vuola, Aarno Oskar; Akram, Saad Ullah; Kannala, Juho.
ISBI 2019 - 2019 IEEE International Symposium on Biomedical Imaging. IEEE, 2019. p. 208-212 8759574 (IEEE International Symposium on Biomedical Imaging).

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

TY - GEN

T1 - Mask-RCNN and u-net ensembled for nuclei segmentation

AU - Vuola, Aarno Oskar

AU - Akram, Saad Ullah

AU - Kannala, Juho

N1 - Conference code: 16

PY - 2019

Y1 - 2019

N2 - Nuclei segmentation is both an important and in some ways ideal task for modern computer vision methods, e.g. convolutional neural networks. While recent developments in theory and open-source software have made these tools easier to implement, expert knowledge is still required to choose the Sight model architecture and training setup. We compare two popular segmentation frameworks, U-Net and Mask-RCNN in the nuclei segmentation task and find that they have different strengths and failures. To get the best of both worlds, we develop an ensemble model to combine their predictions that can outperform both models by a significant margin and should be considered when aiming for best nuclei segmentation performance.

AB - Nuclei segmentation is both an important and in some ways ideal task for modern computer vision methods, e.g. convolutional neural networks. While recent developments in theory and open-source software have made these tools easier to implement, expert knowledge is still required to choose the Sight model architecture and training setup. We compare two popular segmentation frameworks, U-Net and Mask-RCNN in the nuclei segmentation task and find that they have different strengths and failures. To get the best of both worlds, we develop an ensemble model to combine their predictions that can outperform both models by a significant margin and should be considered when aiming for best nuclei segmentation performance.

KW - nuclei segmentation

KW - microscopy image analysis

KW - convolutional neural networks

U2 - 10.1109/ISBI.2019.8759574

DO - 10.1109/ISBI.2019.8759574

M3 - Conference article in proceedings

T3 - IEEE International Symposium on Biomedical Imaging

SP - 208

EP - 212

BT - ISBI 2019 - 2019 IEEE International Symposium on Biomedical Imaging

PB - IEEE

T2 - IEEE International Symposium on Biomedical Imaging

Y2 - 8 April 2019 through 11 April 2019

ER -

Mask-RCNN and u-net ensembled for nuclei segmentation

Abstract

Publication series

Conference

Keywords

Access to Document

Fingerprint

Cite this