TY - GEN
T1 - A Comparative Study of Neural Computing Approaches for Semantic Segmentation of Breast Tumors on Ultrasound Images
AU - Aguilar-Camacho, Luis Eduardo
AU - Gómez-Flores, Wilfrido
AU - Sossa-Azuela, Juan Humberto
N1 - Publisher Copyright:
© 2022, Springer Nature Switzerland AG.
PY - 2022
Y1 - 2022
N2 - This paper compares two approaches for semantic segmentation of breast tumors on ultrasound. The first approach, called conventional, follows the typical pattern classification process to extract hand-crafted features, followed by pixel classification with a Multilayer Perceptron (MLP) network. The second approach, called convolutional, uses a Convolutional Neural Network (CNN) to learn features automatically. For evaluating both approaches, a breast ultrasound dataset with 1200 images is considered. Experimental results reveal that the CNNs called VGG16 and ResNet50 outperformed the conventional approach in various segmentation quality indices. Thus, extracting hand-crafted discriminant features is challenging since it depends on the problem domain and the designer’s skills. On the other hand, through transfer learning, it is possible to adjust a pre-trained CNN for addressing the problem of tumor segmentation satisfactorily. This performance is because CNN learns general features in its first layers, and more subtle features are activated as depth increases.
AB - This paper compares two approaches for semantic segmentation of breast tumors on ultrasound. The first approach, called conventional, follows the typical pattern classification process to extract hand-crafted features, followed by pixel classification with a Multilayer Perceptron (MLP) network. The second approach, called convolutional, uses a Convolutional Neural Network (CNN) to learn features automatically. For evaluating both approaches, a breast ultrasound dataset with 1200 images is considered. Experimental results reveal that the CNNs called VGG16 and ResNet50 outperformed the conventional approach in various segmentation quality indices. Thus, extracting hand-crafted discriminant features is challenging since it depends on the problem domain and the designer’s skills. On the other hand, through transfer learning, it is possible to adjust a pre-trained CNN for addressing the problem of tumor segmentation satisfactorily. This performance is because CNN learns general features in its first layers, and more subtle features are activated as depth increases.
KW - Artificial neural network
KW - Breast ultrasound
KW - Convolutional neural network
KW - Semantic segmentation
UR - http://www.scopus.com/inward/record.url?scp=85123852687&partnerID=8YFLogxK
U2 - 10.1007/978-3-030-70601-2_241
DO - 10.1007/978-3-030-70601-2_241
M3 - Contribución a la conferencia
AN - SCOPUS:85123852687
SN - 9783030706005
T3 - IFMBE Proceedings
SP - 1649
EP - 1657
BT - 27th Brazilian Congress on Biomedical Engineering - Proceedings of CBEB 2020
A2 - Bastos-Filho, Teodiano Freire
A2 - de Oliveira Caldeira, Eliete Maria
A2 - Frizera-Neto, Anselmo
PB - Springer Science and Business Media Deutschland GmbH
T2 - 27th Brazilian Congress on Biomedical Engineering, CBEB 2020
Y2 - 26 October 2020 through 30 October 2020
ER -