RATIONALE AND OBJECTIVES: An artificial neural network (ANN) approach was developed for the computer-aided diagnosis of mammography using an optimally minimized number of input features. METHODS: A backpropagation ANN merged nine input features (age plus eight radiographic findings extracted by radiologists) to predict biopsy outcome as its output. The features were ranked, and more important ones were selected to produce an optimal subset of features. RESULTS: Given all nine features, the ANN performed with a receiver operator characteristic area under the curve (Az) of .95 +/- .01. Given only the four most important features, the ANN performed with an Az of .96 +/- .01. Although not significantly better than the ANN with all nine features, the ANN with the four optimized features was significantly better than expert radiologists' Az of .90 +/- .02 (p = .01). This four-feature ANN had a 95% sensitivity and an 81% specificity. For cases with calcifications, the radiologists' performance dropped to an Az of .85 +/- .04, whereas a specialized three-feature ANN performed significantly better with an Az of .95 +/- .02 (p = .02). CONCLUSION: Given only four input features, the ANN predicted biopsy outcome significantly better than did expert radiologists, who also had access to other radiographic and nonradiographic data. The reduced number of features would substantially decrease data entry efforts and potentially improve the ANN's general applicability.
RATIONALE AND OBJECTIVES: An artificial neural network (ANN) approach was developed for the computer-aided diagnosis of mammography using an optimally minimized number of input features. METHODS: A backpropagation ANN merged nine input features (age plus eight radiographic findings extracted by radiologists) to predict biopsy outcome as its output. The features were ranked, and more important ones were selected to produce an optimal subset of features. RESULTS: Given all nine features, the ANN performed with a receiver operator characteristic area under the curve (Az) of .95 +/- .01. Given only the four most important features, the ANN performed with an Az of .96 +/- .01. Although not significantly better than the ANN with all nine features, the ANN with the four optimized features was significantly better than expert radiologists' Az of .90 +/- .02 (p = .01). This four-feature ANN had a 95% sensitivity and an 81% specificity. For cases with calcifications, the radiologists' performance dropped to an Az of .85 +/- .04, whereas a specialized three-feature ANN performed significantly better with an Az of .95 +/- .02 (p = .02). CONCLUSION: Given only four input features, the ANN predicted biopsy outcome significantly better than did expert radiologists, who also had access to other radiographic and nonradiographic data. The reduced number of features would substantially decrease data entry efforts and potentially improve the ANN's general applicability.
Authors: Turgay Ayer; Oguzhan Alagoz; Jagpreet Chhatwal; Jude W Shavlik; Charles E Kahn; Elizabeth S Burnside Journal: Cancer Date: 2010-07-15 Impact factor: 6.860
Authors: Hui Li; Maryellen L Giger; Yading Yuan; Weijie Chen; Karla Horsch; Li Lan; Andrew R Jamieson; Charlene A Sennett; Sanaz A Jansen Journal: Acad Radiol Date: 2008-11 Impact factor: 3.173