Determinação de um conjunto reduzido de características morfológicas para classificação de lesões mamárias em imagens de ultrassom utilizando máquinas de vetores de suporte

• Carmina Nascimento CETELI/UFAM
• Thales Silva CETELI/UFAM
• Wagner Pereira CETELI/UFAM
• Marly Costa CETELI/UFAM
• Cícero C. Filho CETELI/UFAM

DOI: https://doi.org/10.5753/sbcas.2015.10361

Abstract

The use of tools for computer-aided diagnosis (CAD) has been proposed for the detection and classification of breast cancer. Concerning breast cancer image diagnosis with ultrasound, some results found in literature show a better performance of morphological features on breast cancer lesion differentiating and that a reduced set of features shows a better performance than a large set of features. In this study, we evaluated the performance SVM classifiers with different kernels: polynomial and RBF. Different sets of morphological features were used for SVM training and classification. To select reduced sets of features, scalar selection technique with correlation was used. The best results obtained for accuracy and area under ROC curve were 92% and 0.920, respectively.

References

Cheng, H. D. et al. Automated breast cancer detection and classification using ultrasound images: A survey. Pattern Recognition, v. 43, n. 1, p. 299-317, 2010.

Flores, W. G. Desarrollo de una Metodología Computacional para la Clasificación de Lesiones de Mama en Imágenes Ultrasónicas. 2009. Departamento de Ingeniería Eléctrica Sección de Bioelectrónica, Instituto Politécnico Nacional

Flores, W. G.; Pereira, W. C. D. A.; Infantosi, A. F. C. Improving classification performance of breast lesions on ultrasonography. Pattern Recognition, v. 48, n. 4, p. 1125 - 1136, 2015.

Haykin, S. S. Neural Networks and Learning Machines. 2009.

Huang, Y. L.; Wang, K. L.; Chen, D. R. Diagnosis of breast tumors with ultrasonic texture analysis using support vector machines. Neural Computing & Applications, v. 15, n. 2, p. 164-169, 2006.

INCA. Mama. p. Câncer de Mama, 2015. Disponível em: (http://www2.inca.gov.br/wps/wcm/connect/tiposdecancer/site/home/mama/cancer_ mama .

Kohavi, R. A Study of Cross-Validation and Bootstrap for Accuracy Estimation and Model Selection. International Joint Conference on Artificial Intelligence, v. 2, p. 1137-1143, 1995.

Pereira, W. C. A. et al. A non-linear morphometric feature selection approach for breast tumor contour from ultrasonic images. Computers in Biology and Medicine, 2010. SBCANCEROLOGIA. Cancer de Mama. 2015. Disponível em: http://www.sbcancer.org.br/home2/site/index.php?option=com_content&view=articl e&id=110:cancer-de-mama&catid=29&Itemid=123 . Acesso em: 02/12/2015.

Skaane, P. Ultrasonography as adjunct to mammography in the evaluation of breast tumors. Actaradiologica. Supplementum, 1999

Theodoridis, S.; Koutroumbas, K. Pattern Recognition, Fourth Edition. 4th. Academic Press, 2008.

Uniyal, N. et al. A new approach to ultrasonic detection of malignant breast tumors. 2013 Ieee International Ultrasonics Symposium (Ius), p. 96-99, 2013

WHO. Breast cancer: prevention and control. 2014. Disponível em: http://www.who.int/cancer/detection/breastcancer/en/index4.html.

Wu, W.-J.; Moon, W. K. Ultrasound breast tumor image computer-aided diagnosis with texture and morphological features. Academic Radiology, v. 15, n.7, p. 873-880, 2008.

Yeh, C.-K. et al. Automatic ultrasonic breast lesions detection using support vector machine based algorithm. Medical Imaging: Ultrasonic Imaging and Signal Processing, p.51317-51317, 2007.

Zhou, S. et al. Shearlet-based texture feature extraction for classification of breast tumor in ultrasound image. Biomedical Signal Processing and Control, v. 8, n. 6, p. 688-696, 2013.

Zyout, I.; Czajkowska, J., Marcin. Multi-scale Textural Feature Extraction and Particle Swarm Optimization Based Model Selection for False Positive Reduction in Mammography. Computerized Medical Imaging and Graphics, 2015.