Eliminating false-positive microcalcification clusters in a mammography CAD scheme using a Bayesian neural network

D. C. Edwards, J. Papaioannou, Y. Jiang, Matthew A Kupinski, R. M. Nishikawa

Research output: Chapter in Book/Report/Conference proceedingConference contribution

7 Citations (Scopus)

Abstract

We have applied a Bayesian neural network (BNN) to the task of distinguishing between true-positive (TP) and false-positive (FP) detected clusters in a computer-aided diagnosis (CAD) scheme for detecting clustered microcalcifications in mammograms. Because BNNs can approximate ideal observer decision functions given sufficient training data, this approach should have better performance than our previous FP cluster elimination methods. Eight cluster-based features were extracted from the TP and FP clusters detected by the scheme in a training dataset of 39 mammograms. This set of features was used to train a BNN with eight input nodes, five hidden nodes, and one output node. The trained BNN was tested on the TP and FP clusters detected by our scheme in an independent testing set of 50 mammograms. The BNN output was analyzed using ROC and FROC analysis. The detection scheme with the BNN for FP cluster elimination had substantially better cluster sensitivity at low FP rates (below 0.8 FP clusters per image) than the original detection scheme without the BNN. Our preliminary research shows that a BNN can improve the performance of our scheme for detecting clusters of microcalcifications.

Original languageEnglish (US)
Title of host publicationProceedings of SPIE - The International Society for Optical Engineering
EditorsM. Sonka, K.M. Hanson
Pages1954-1960
Number of pages7
Volume4322
Edition3
DOIs
StatePublished - 2001
Externally publishedYes
EventMedical Imaging 2001 Image Processing - San Diego, CA, United States
Duration: Feb 19 2001Feb 22 2001

Other

OtherMedical Imaging 2001 Image Processing
CountryUnited States
CitySan Diego, CA
Period2/19/012/22/01

Fingerprint

Computer aided diagnosis
Mammography
Neural networks
elimination
education
output
Testing

Keywords

  • Bayesian artificial neural networks
  • CAD
  • Clustered microcalcifications
  • Ideal observer approximation
  • Mammography

ASJC Scopus subject areas

  • Electrical and Electronic Engineering
  • Condensed Matter Physics

Cite this

Edwards, D. C., Papaioannou, J., Jiang, Y., Kupinski, M. A., & Nishikawa, R. M. (2001). Eliminating false-positive microcalcification clusters in a mammography CAD scheme using a Bayesian neural network. In M. Sonka, & K. M. Hanson (Eds.), Proceedings of SPIE - The International Society for Optical Engineering (3 ed., Vol. 4322, pp. 1954-1960) https://doi.org/10.1117/12.431089

Eliminating false-positive microcalcification clusters in a mammography CAD scheme using a Bayesian neural network. / Edwards, D. C.; Papaioannou, J.; Jiang, Y.; Kupinski, Matthew A; Nishikawa, R. M.

Proceedings of SPIE - The International Society for Optical Engineering. ed. / M. Sonka; K.M. Hanson. Vol. 4322 3. ed. 2001. p. 1954-1960.

Research output: Chapter in Book/Report/Conference proceedingConference contribution

Edwards, DC, Papaioannou, J, Jiang, Y, Kupinski, MA & Nishikawa, RM 2001, Eliminating false-positive microcalcification clusters in a mammography CAD scheme using a Bayesian neural network. in M Sonka & KM Hanson (eds), Proceedings of SPIE - The International Society for Optical Engineering. 3 edn, vol. 4322, pp. 1954-1960, Medical Imaging 2001 Image Processing, San Diego, CA, United States, 2/19/01. https://doi.org/10.1117/12.431089
Edwards DC, Papaioannou J, Jiang Y, Kupinski MA, Nishikawa RM. Eliminating false-positive microcalcification clusters in a mammography CAD scheme using a Bayesian neural network. In Sonka M, Hanson KM, editors, Proceedings of SPIE - The International Society for Optical Engineering. 3 ed. Vol. 4322. 2001. p. 1954-1960 https://doi.org/10.1117/12.431089
Edwards, D. C. ; Papaioannou, J. ; Jiang, Y. ; Kupinski, Matthew A ; Nishikawa, R. M. / Eliminating false-positive microcalcification clusters in a mammography CAD scheme using a Bayesian neural network. Proceedings of SPIE - The International Society for Optical Engineering. editor / M. Sonka ; K.M. Hanson. Vol. 4322 3. ed. 2001. pp. 1954-1960
@inproceedings{1640f0d0b7b3471eae51c168d72accdb,
title = "Eliminating false-positive microcalcification clusters in a mammography CAD scheme using a Bayesian neural network",
abstract = "We have applied a Bayesian neural network (BNN) to the task of distinguishing between true-positive (TP) and false-positive (FP) detected clusters in a computer-aided diagnosis (CAD) scheme for detecting clustered microcalcifications in mammograms. Because BNNs can approximate ideal observer decision functions given sufficient training data, this approach should have better performance than our previous FP cluster elimination methods. Eight cluster-based features were extracted from the TP and FP clusters detected by the scheme in a training dataset of 39 mammograms. This set of features was used to train a BNN with eight input nodes, five hidden nodes, and one output node. The trained BNN was tested on the TP and FP clusters detected by our scheme in an independent testing set of 50 mammograms. The BNN output was analyzed using ROC and FROC analysis. The detection scheme with the BNN for FP cluster elimination had substantially better cluster sensitivity at low FP rates (below 0.8 FP clusters per image) than the original detection scheme without the BNN. Our preliminary research shows that a BNN can improve the performance of our scheme for detecting clusters of microcalcifications.",
keywords = "Bayesian artificial neural networks, CAD, Clustered microcalcifications, Ideal observer approximation, Mammography",
author = "Edwards, {D. C.} and J. Papaioannou and Y. Jiang and Kupinski, {Matthew A} and Nishikawa, {R. M.}",
year = "2001",
doi = "10.1117/12.431089",
language = "English (US)",
volume = "4322",
pages = "1954--1960",
editor = "M. Sonka and K.M. Hanson",
booktitle = "Proceedings of SPIE - The International Society for Optical Engineering",
edition = "3",

}

TY - GEN

T1 - Eliminating false-positive microcalcification clusters in a mammography CAD scheme using a Bayesian neural network

AU - Edwards, D. C.

AU - Papaioannou, J.

AU - Jiang, Y.

AU - Kupinski, Matthew A

AU - Nishikawa, R. M.

PY - 2001

Y1 - 2001

N2 - We have applied a Bayesian neural network (BNN) to the task of distinguishing between true-positive (TP) and false-positive (FP) detected clusters in a computer-aided diagnosis (CAD) scheme for detecting clustered microcalcifications in mammograms. Because BNNs can approximate ideal observer decision functions given sufficient training data, this approach should have better performance than our previous FP cluster elimination methods. Eight cluster-based features were extracted from the TP and FP clusters detected by the scheme in a training dataset of 39 mammograms. This set of features was used to train a BNN with eight input nodes, five hidden nodes, and one output node. The trained BNN was tested on the TP and FP clusters detected by our scheme in an independent testing set of 50 mammograms. The BNN output was analyzed using ROC and FROC analysis. The detection scheme with the BNN for FP cluster elimination had substantially better cluster sensitivity at low FP rates (below 0.8 FP clusters per image) than the original detection scheme without the BNN. Our preliminary research shows that a BNN can improve the performance of our scheme for detecting clusters of microcalcifications.

AB - We have applied a Bayesian neural network (BNN) to the task of distinguishing between true-positive (TP) and false-positive (FP) detected clusters in a computer-aided diagnosis (CAD) scheme for detecting clustered microcalcifications in mammograms. Because BNNs can approximate ideal observer decision functions given sufficient training data, this approach should have better performance than our previous FP cluster elimination methods. Eight cluster-based features were extracted from the TP and FP clusters detected by the scheme in a training dataset of 39 mammograms. This set of features was used to train a BNN with eight input nodes, five hidden nodes, and one output node. The trained BNN was tested on the TP and FP clusters detected by our scheme in an independent testing set of 50 mammograms. The BNN output was analyzed using ROC and FROC analysis. The detection scheme with the BNN for FP cluster elimination had substantially better cluster sensitivity at low FP rates (below 0.8 FP clusters per image) than the original detection scheme without the BNN. Our preliminary research shows that a BNN can improve the performance of our scheme for detecting clusters of microcalcifications.

KW - Bayesian artificial neural networks

KW - CAD

KW - Clustered microcalcifications

KW - Ideal observer approximation

KW - Mammography

UR - http://www.scopus.com/inward/record.url?scp=0034855208&partnerID=8YFLogxK

UR - http://www.scopus.com/inward/citedby.url?scp=0034855208&partnerID=8YFLogxK

U2 - 10.1117/12.431089

DO - 10.1117/12.431089

M3 - Conference contribution

AN - SCOPUS:0034855208

VL - 4322

SP - 1954

EP - 1960

BT - Proceedings of SPIE - The International Society for Optical Engineering

A2 - Sonka, M.

A2 - Hanson, K.M.

ER -