Skip to main content

Lund University Publications

LUND UNIVERSITY LIBRARIES

Ensemble of Deep Convolutional Neural Networks with Monte Carlo Dropout Sampling for Automated Image Segmentation Quality Control and Robust Deep Learning Using Small Datasets

Hann, Evan ; Gonzales, Ricardo A. LU orcid ; Popescu, Iulia A. ; Zhang, Qiang ; Ferreira, Vanessa M. and Piechnik, Stefan K. (2021) 25th Annual Conference on Medical Image Understanding and Analysis, MIUA 2021 In Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics) 12722 LNCS. p.280-293
Abstract

Recent progress on deep learning (DL)-based medical image segmentation can enable fast extraction of clinical parameters for efficient clinical workflows. However, current DL methods can still fail and require manual visual inspection of outputs, which is time-consuming and diminishes the advantages of automation. For clinical applications, it is essential to develop DL approaches that can not only perform accurate segmentation, but also predict the segmentation quality and flag poor-quality results to avoid errors in diagnosis. To achieve robust performance, DL-based methods often require large datasets, which are not always readily available. It would be highly desirable to be able to train DL models using only small datasets, but... (More)

Recent progress on deep learning (DL)-based medical image segmentation can enable fast extraction of clinical parameters for efficient clinical workflows. However, current DL methods can still fail and require manual visual inspection of outputs, which is time-consuming and diminishes the advantages of automation. For clinical applications, it is essential to develop DL approaches that can not only perform accurate segmentation, but also predict the segmentation quality and flag poor-quality results to avoid errors in diagnosis. To achieve robust performance, DL-based methods often require large datasets, which are not always readily available. It would be highly desirable to be able to train DL models using only small datasets, but this requires a quality prediction method to ensure reliability. We present a novel segmentation framework utilizing an ensemble of deep convolutional neural networks with Monte Carlo sampling. The proposed framework merges the advantages of both state-of-the-art deep ensembles and Bayesian approaches, to provide robust segmentation with inherent quality control. We successfully developed and tested this framework using just a small MRI dataset of 45 subjects. The framework obtained high mean Dice similarity coefficients (DSC) for segmentation of the endocardium (0.922) and the epicardium (0.942); importantly, segmentation DSC can be accurately predicted with low mean absolute errors (≤0.035), in the absence of the manual ground truth. Furthermore, binary classification of segmentation quality achieved a near-perfect accuracy of 99%. The proposed framework can enable fast and reliable medical image analysis with accurate quality control, and training of DL-based methods using even small datasets.

(Less)
Please use this url to cite or link to this publication:
author
; ; ; ; and
publishing date
type
Chapter in Book/Report/Conference proceeding
publication status
published
subject
keywords
Automated quality assessment, Ensemble learning, Monte Carlo sampling, Segmentation
host publication
Medical Image Understanding and Analysis - 25th Annual Conference, MIUA 2021, Proceedings
series title
Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)
editor
Papież, Bartłomiej W. ; Yaqub, Mohammad ; Jiao, Jianbo ; Namburete, Ana I. and Noble, J. Alison
volume
12722 LNCS
pages
280 - 293
publisher
Springer Science and Business Media B.V.
conference name
25th Annual Conference on Medical Image Understanding and Analysis, MIUA 2021
conference location
Virtual, Online
conference dates
2021-07-12 - 2021-07-14
external identifiers
  • scopus:85112221543
ISSN
0302-9743
1611-3349
ISBN
9783030804312
DOI
10.1007/978-3-030-80432-9_22
language
English
LU publication?
no
additional info
Publisher Copyright: © 2021, Springer Nature Switzerland AG.
id
2ceecef9-91b6-4aca-9eeb-4d674047089f
date added to LUP
2021-10-28 13:38:19
date last changed
2024-06-15 19:22:55
@inproceedings{2ceecef9-91b6-4aca-9eeb-4d674047089f,
  abstract     = {{<p>Recent progress on deep learning (DL)-based medical image segmentation can enable fast extraction of clinical parameters for efficient clinical workflows. However, current DL methods can still fail and require manual visual inspection of outputs, which is time-consuming and diminishes the advantages of automation. For clinical applications, it is essential to develop DL approaches that can not only perform accurate segmentation, but also predict the segmentation quality and flag poor-quality results to avoid errors in diagnosis. To achieve robust performance, DL-based methods often require large datasets, which are not always readily available. It would be highly desirable to be able to train DL models using only small datasets, but this requires a quality prediction method to ensure reliability. We present a novel segmentation framework utilizing an ensemble of deep convolutional neural networks with Monte Carlo sampling. The proposed framework merges the advantages of both state-of-the-art deep ensembles and Bayesian approaches, to provide robust segmentation with inherent quality control. We successfully developed and tested this framework using just a small MRI dataset of 45 subjects. The framework obtained high mean Dice similarity coefficients (DSC) for segmentation of the endocardium (0.922) and the epicardium (0.942); importantly, segmentation DSC can be accurately predicted with low mean absolute errors (≤0.035), in the absence of the manual ground truth. Furthermore, binary classification of segmentation quality achieved a near-perfect accuracy of 99%. The proposed framework can enable fast and reliable medical image analysis with accurate quality control, and training of DL-based methods using even small datasets.</p>}},
  author       = {{Hann, Evan and Gonzales, Ricardo A. and Popescu, Iulia A. and Zhang, Qiang and Ferreira, Vanessa M. and Piechnik, Stefan K.}},
  booktitle    = {{Medical Image Understanding and Analysis - 25th Annual Conference, MIUA 2021, Proceedings}},
  editor       = {{Papież, Bartłomiej W. and Yaqub, Mohammad and Jiao, Jianbo and Namburete, Ana I. and Noble, J. Alison}},
  isbn         = {{9783030804312}},
  issn         = {{0302-9743}},
  keywords     = {{Automated quality assessment; Ensemble learning; Monte Carlo sampling; Segmentation}},
  language     = {{eng}},
  pages        = {{280--293}},
  publisher    = {{Springer Science and Business Media B.V.}},
  series       = {{Lecture Notes in Computer Science (including subseries Lecture Notes in Artificial Intelligence and Lecture Notes in Bioinformatics)}},
  title        = {{Ensemble of Deep Convolutional Neural Networks with Monte Carlo Dropout Sampling for Automated Image Segmentation Quality Control and Robust Deep Learning Using Small Datasets}},
  url          = {{http://dx.doi.org/10.1007/978-3-030-80432-9_22}},
  doi          = {{10.1007/978-3-030-80432-9_22}},
  volume       = {{12722 LNCS}},
  year         = {{2021}},
}