Skip to main content

Lund University Publications

LUND UNIVERSITY LIBRARIES

Using Bayesian neural networks with ARD input selection to detect malignant ovarian masses prior to surgery

Van Calster, Ben ; Timmerman, Dirk ; Nabney, Ian T. ; Valentin, Lil LU orcid ; Testa, Antonia C. ; Van Holsbeke, Caroline ; Vergote, Ignace and Van Huffel, Sabine (2008) 28th Annual International Conference of the IEEE-Engineering-in-Medicine-and-Biology-Society 17(5-6). p.489-500
Abstract
In this paper, we applied Bayesian multi-layer perceptrons (MLP) using the evidence procedure to predict malignancy of ovarian masses in a large (n = 1,066) multi-centre data set. Automatic relevance determination (ARD) was used to select the most relevant inputs. Fivefold cross-validation (5CV) and repeated 5CV was used to select the optimal combination of input set and number of hidden neurons. Results indicate good performance of the models with area under the receiver operating characteristic curve values of 0.93-0.94 on independent data. Comparison with a linear benchmark model and a previously developed logistic regression model shows that the present problem is very well linearly separable. A resampling analysis further shows that... (More)
In this paper, we applied Bayesian multi-layer perceptrons (MLP) using the evidence procedure to predict malignancy of ovarian masses in a large (n = 1,066) multi-centre data set. Automatic relevance determination (ARD) was used to select the most relevant inputs. Fivefold cross-validation (5CV) and repeated 5CV was used to select the optimal combination of input set and number of hidden neurons. Results indicate good performance of the models with area under the receiver operating characteristic curve values of 0.93-0.94 on independent data. Comparison with a linear benchmark model and a previously developed logistic regression model shows that the present problem is very well linearly separable. A resampling analysis further shows that the number of hidden neurons specified in the ARD analyses for input selection may influence model performance. This paper shows that Bayesian MLPs, although not frequently used, are a useful tool for detecting malignant ovarian tumours. (Less)
Please use this url to cite or link to this publication:
author
; ; ; ; ; ; and
organization
publishing date
type
Chapter in Book/Report/Conference proceeding
publication status
published
subject
keywords
ultrasound, automatic relevance determination, netlab, evidence framework, Bayesian, ovarian tumour classification, multi-layer perceptrons
host publication
NEURAL COMPUTING & APPLICATIONS
volume
17
issue
5-6
pages
489 - 500
publisher
Springer
conference name
28th Annual International Conference of the IEEE-Engineering-in-Medicine-and-Biology-Society
conference dates
2006-08-30 - 2006-09-03
external identifiers
  • wos:000259483200007
  • scopus:52349091540
ISSN
0941-0643
DOI
10.1007/s00521-007-0147-1
language
English
LU publication?
yes
id
2e8b93df-2207-4bd1-8873-fbbe9de5fa43 (old id 1286605)
date added to LUP
2016-04-01 14:26:06
date last changed
2022-03-22 00:01:49
@inproceedings{2e8b93df-2207-4bd1-8873-fbbe9de5fa43,
  abstract     = {{In this paper, we applied Bayesian multi-layer perceptrons (MLP) using the evidence procedure to predict malignancy of ovarian masses in a large (n = 1,066) multi-centre data set. Automatic relevance determination (ARD) was used to select the most relevant inputs. Fivefold cross-validation (5CV) and repeated 5CV was used to select the optimal combination of input set and number of hidden neurons. Results indicate good performance of the models with area under the receiver operating characteristic curve values of 0.93-0.94 on independent data. Comparison with a linear benchmark model and a previously developed logistic regression model shows that the present problem is very well linearly separable. A resampling analysis further shows that the number of hidden neurons specified in the ARD analyses for input selection may influence model performance. This paper shows that Bayesian MLPs, although not frequently used, are a useful tool for detecting malignant ovarian tumours.}},
  author       = {{Van Calster, Ben and Timmerman, Dirk and Nabney, Ian T. and Valentin, Lil and Testa, Antonia C. and Van Holsbeke, Caroline and Vergote, Ignace and Van Huffel, Sabine}},
  booktitle    = {{NEURAL COMPUTING & APPLICATIONS}},
  issn         = {{0941-0643}},
  keywords     = {{ultrasound; automatic relevance determination; netlab; evidence framework; Bayesian; ovarian tumour classification; multi-layer perceptrons}},
  language     = {{eng}},
  number       = {{5-6}},
  pages        = {{489--500}},
  publisher    = {{Springer}},
  title        = {{Using Bayesian neural networks with ARD input selection to detect malignant ovarian masses prior to surgery}},
  url          = {{http://dx.doi.org/10.1007/s00521-007-0147-1}},
  doi          = {{10.1007/s00521-007-0147-1}},
  volume       = {{17}},
  year         = {{2008}},
}