Impact of missing data in evaluating artificial neural networks trained on complete data

Mia K. Markey; Georgia D. Tourassi; Michael Margolis; David M. DeLong

doi:10.1016/j.compbiomed.2005.02.001

Impact of missing data in evaluating artificial neural networks trained on complete data

Mia K. Markey, Georgia D. Tourassi, Michael Margolis, David M. DeLong

Research output: Contribution to journal › Article › peer-review

40 Scopus citations

Abstract

This study investigated the impact of missing data in the evaluation of artificial neural network (ANN) models trained on complete data for the task of predicting whether breast lesions are benign or malignant from their mammographic Breast Imaging and Reporting Data System (BI- RADS ^TM) descriptors. A feed-forward, back-propagation ANN was tested with three methods for estimating the missing values. Similar results were achieved with a constraint satisfaction ANN, which can accommodate missing values without a separate estimation step. This empirical study highlights the need for additional research on developing robust clinical decision support systems for realistic environments in which key information may be unknown or inaccessible.

Original language	English
Pages (from-to)	516-525
Number of pages	10
Journal	Computers in Biology and Medicine
Volume	36
Issue number	5
DOIs	https://doi.org/10.1016/j.compbiomed.2005.02.001
State	Published - May 2006
Externally published	Yes

Funding

Georgia Tourassi , is an Assistant Research Professor in the Department of Radiology at Duke University Medical Center and Adjunct Associate Professor of Computer Engineering and Computer Science at the University of Louisville. She earned a B.S. in Physics from the University of Thessaloniki in Greece and the Ph.D. in Biomedical Engineering from Duke University. Dr. Tourassi is a member of the Institute of Electrical and Electronics Engineers (IEEE), the International Society for Optical Engineering (SPIE), the American Association of Physicists in Medicine (AAPM), and the Radiological Society of North America (RSNA). She is the associate editor in Radiology and referee in several journals such as Medical Physics, Academic Radiology, and Journal of Electronic Imaging. Her research interests include applications of artificial intelligence in computer-aided medical diagnosis and medical image processing where she has numerous publications. Her research work has been supported by the National Institutes of Health, US, Army, and the Whitaker Foundation.

Keywords

Breast neoplasms
Computer-assisted
Diagnosis
Mammography

Access to Document

10.1016/j.compbiomed.2005.02.001

Cite this

@article{f7034d08d0c94afba063c2e1ec55a04a,

title = "Impact of missing data in evaluating artificial neural networks trained on complete data",

abstract = "This study investigated the impact of missing data in the evaluation of artificial neural network (ANN) models trained on complete data for the task of predicting whether breast lesions are benign or malignant from their mammographic Breast Imaging and Reporting Data System (BI- RADS TM) descriptors. A feed-forward, back-propagation ANN was tested with three methods for estimating the missing values. Similar results were achieved with a constraint satisfaction ANN, which can accommodate missing values without a separate estimation step. This empirical study highlights the need for additional research on developing robust clinical decision support systems for realistic environments in which key information may be unknown or inaccessible.",

keywords = "Breast neoplasms, Computer-assisted, Diagnosis, Mammography",

author = "Markey, {Mia K.} and Tourassi, {Georgia D.} and Michael Margolis and DeLong, {David M.}",

year = "2006",

month = may,

doi = "10.1016/j.compbiomed.2005.02.001",

language = "English",

volume = "36",

pages = "516--525",

journal = "Computers in Biology and Medicine",

issn = "0010-4825",

publisher = "Elsevier",

number = "5",

}

TY - JOUR

T1 - Impact of missing data in evaluating artificial neural networks trained on complete data

AU - Markey, Mia K.

AU - Tourassi, Georgia D.

AU - Margolis, Michael

AU - DeLong, David M.

PY - 2006/5

Y1 - 2006/5

N2 - This study investigated the impact of missing data in the evaluation of artificial neural network (ANN) models trained on complete data for the task of predicting whether breast lesions are benign or malignant from their mammographic Breast Imaging and Reporting Data System (BI- RADS TM) descriptors. A feed-forward, back-propagation ANN was tested with three methods for estimating the missing values. Similar results were achieved with a constraint satisfaction ANN, which can accommodate missing values without a separate estimation step. This empirical study highlights the need for additional research on developing robust clinical decision support systems for realistic environments in which key information may be unknown or inaccessible.

AB - This study investigated the impact of missing data in the evaluation of artificial neural network (ANN) models trained on complete data for the task of predicting whether breast lesions are benign or malignant from their mammographic Breast Imaging and Reporting Data System (BI- RADS TM) descriptors. A feed-forward, back-propagation ANN was tested with three methods for estimating the missing values. Similar results were achieved with a constraint satisfaction ANN, which can accommodate missing values without a separate estimation step. This empirical study highlights the need for additional research on developing robust clinical decision support systems for realistic environments in which key information may be unknown or inaccessible.

KW - Breast neoplasms

KW - Computer-assisted

KW - Diagnosis

KW - Mammography

UR - http://www.scopus.com/inward/record.url?scp=33646470217&partnerID=8YFLogxK

U2 - 10.1016/j.compbiomed.2005.02.001

DO - 10.1016/j.compbiomed.2005.02.001

M3 - Article

C2 - 15893745

AN - SCOPUS:33646470217

SN - 0010-4825

VL - 36

SP - 516

EP - 525

JO - Computers in Biology and Medicine

JF - Computers in Biology and Medicine

IS - 5

ER -

Impact of missing data in evaluating artificial neural networks trained on complete data

Abstract

Funding

Keywords

Access to Document

Other files and links

Fingerprint

Cite this