Chen et al. (1999)
Chen D-R, Chang R-F, Huang Y-L. Computer-aided diagnosis applied to us of solid breast nodules by using neural networks. Radiology. 1999; 213(2):407-12. PMid:10551220. http://dx.doi.org/10.1148/radiology.213.2.r99nv13407. http://dx.doi.org/10.1148/radiology.213....
|
Feedforward Neural Network, backpropagation algorithm, 25-10-1, 10 cross-validation, stop criteria: mean square error. |
144 sonograms. 52 malignant and 88 benign tumors. |
Textural features: 5×5 2D-autocorrelation matrix. |
Acc: 95%; Sens: 98%; Spec: 93%; Mean AUC: 0.731 ± 0.040 (SD). |
Chen et al. (2003)
Chen C-M, Chou Y-H, Han K-C, Hung G-S, Tiu C-M, Chiou H-J, Chiou SY. Breast lesions on sonograms: computer-aided diagnosis with nearly setting-independent features and artificial neural networks. Radiology. 2003; 226(2):504-14. PMid:12563146. http://dx.doi.org/10.1148/radiol.2262011843. http://dx.doi.org/10.1148/radiol.2262011...
|
Feedforward Neural Network, backpropagation algorithm, 7-10-1, one-leave-out training-test methodology, stop criteria: mean square error. |
271 sonograms, 140 malignant and 131 benign tumors. |
7 morphological features. |
Acc: 92.8%; Sens: 96.7%; Spec: 87.7%; Mean AUC: 0.952 ± 0.018 (SD). |
Alvarenga et al. (2005)
Alvarenga AV, Pereira WCA, Infantosi AFC, Azevedo CM. Classification of breast tumours on ultrasound images using morphometric parameters. In: Ruano MG, Ruano AE, editors. Proceedings of the International Workshop on Intelligent Signal Processing; 2005 Sept 1-3; Algarve, Portugal. Piscataway: IEEE; 2005. p. 206-10.
|
Feedforward Neural Network, GA-Back propagation Hybrid Training. |
152 sonograms, 100 malignant and 52 benign tumors. |
6 morphological parameters; Convex polygon parameters; circularity. |
Acc: 90%; Sens: 90%; Spec: 90%, PPV: 93.7%; NPV: 84.4%. |
Wu and Moon (2008)
Wu W-J, Moon WK. Ultrasound breast tumor image computer-aided diagnosis with texture and morphological features. Academic Radiology. 2008; 15(7):873-80. PMid:18572123. http://dx.doi.org/10.1016/j.acra.2008.01.010. http://dx.doi.org/10.1016/j.acra.2008.01...
|
SVM, 5-fold-cross-validation, kernel: non-linear gaussian basis. |
210 sonograms, 100 malignant and 120 benign tumors. |
Autocovariance texture features and solidity morphologic features. |
Acc: 92.86%; Sens: 94.44%; Spec: 91.67%; Maximum AUC: 0.949. |
Huang et al. (2008)
Huang YL, Chen DR, Jiang YR, Kuo SJ, Wu HK, Moon WK. Computer-aided diagnosis using morphological features for classifying breast lesions on ultrasound. Ultrasound in Obstetrics & Gynecology. 2008; 32(4):565-72. PMid:18383556. http://dx.doi.org/10.1002/uog.5205. http://dx.doi.org/10.1002/uog.5205...
|
SVM, 10-fold-cross-validation with all characteristics. Feature selection with PCA. |
118 sonograms, 34 malignant and 84 benign lesions. |
19 morphological features. |
Acc: 82.8%; Sen: 94.1%; Spec: 77.3%; Mean AUC: 0.886 ± 0.031 (SD). |
Renjie et al. (2011)
Renjie L, Tao W, Zengchang Q. Classification of benign and malignant breast tumors in ultrasound images based on multiple sonographic and textural features. In: International Conference on Intelligent Human-Machine Systems and Cybernetics (IHMSC 2011); 2011 Aug 26-27; Hangzhou, China. Los Alamitos: IEEE Computer Society; 2011. p. 71-4.
|
SVM, kernels: linear, polynomial, gaussian radial basis function and sigmoid function. |
321 sonograms. |
Sonographic features; Texture features based on SGLD matrix. |
Acc: 86.92%; Sens: 75.18%; Spec: 96.11%. |