Evaluating the performance of convolutional neural networks with direct acyclic graph architectures in automatic segmentation of breast lesion in US images
Abstract Background Outlining lesion contours in Ultra Sound (US) breast images is an important step in breast cancer diagnosis. Malignant lesions infiltrate the surrounding tissue, generating irregular contours, with spiculation and angulated margins, whereas benign lesions produce contours with a...
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
BMC
2019-11-01
|
| Series: | BMC Medical Imaging |
| Subjects: | |
| Online Access: | http://link.springer.com/article/10.1186/s12880-019-0389-2 |
| _version_ | 1819207165944004608 |
|---|---|
| author | Marly Guimarães Fernandes Costa João Paulo Mendes Campos Gustavo de Aquino e Aquino Wagner Coelho de Albuquerque Pereira Cícero Ferreira Fernandes Costa Filho |
| author_facet | Marly Guimarães Fernandes Costa João Paulo Mendes Campos Gustavo de Aquino e Aquino Wagner Coelho de Albuquerque Pereira Cícero Ferreira Fernandes Costa Filho |
| author_sort | Marly Guimarães Fernandes Costa |
| collection | DOAJ |
| description | Abstract Background Outlining lesion contours in Ultra Sound (US) breast images is an important step in breast cancer diagnosis. Malignant lesions infiltrate the surrounding tissue, generating irregular contours, with spiculation and angulated margins, whereas benign lesions produce contours with a smooth outline and elliptical shape. In breast imaging, the majority of the existing publications in the literature focus on using Convolutional Neural Networks (CNNs) for segmentation and classification of lesions in mammographic images. In this study our main objective is to assess the ability of CNNs in detecting contour irregularities in breast lesions in US images. Methods In this study we compare the performance of two CNNs with Direct Acyclic Graph (DAG) architecture and one CNN with a series architecture for breast lesion segmentation in US images. DAG and series architectures are both feedforward networks. The difference is that a DAG architecture could have more than one path between the first layer and end layer, whereas a series architecture has only one path from the beginning layer to the end layer. The CNN architectures were evaluated with two datasets. Results With the more complex DAG architecture, the following mean values were obtained for the metrics used to evaluate the segmented contours: global accuracy: 0.956; IOU: 0.876; F measure: 68.77%; Dice coefficient: 0.892. Conclusion The CNN DAG architecture shows the best metric values used for quantitatively evaluating the segmented contours compared with the gold-standard contours. The segmented contours obtained with this architecture also have more details and irregularities, like the gold-standard contours. |
| first_indexed | 2024-12-23T05:19:10Z |
| format | Article |
| id | doaj.art-df360846781d45e9a7898e98c2a9c395 |
| institution | Directory Open Access Journal |
| issn | 1471-2342 |
| language | English |
| last_indexed | 2024-12-23T05:19:10Z |
| publishDate | 2019-11-01 |
| publisher | BMC |
| record_format | Article |
| series | BMC Medical Imaging |
| spelling | doaj.art-df360846781d45e9a7898e98c2a9c3952022-12-21T17:58:45ZengBMCBMC Medical Imaging1471-23422019-11-0119111310.1186/s12880-019-0389-2Evaluating the performance of convolutional neural networks with direct acyclic graph architectures in automatic segmentation of breast lesion in US imagesMarly Guimarães Fernandes Costa0João Paulo Mendes Campos1Gustavo de Aquino e Aquino2Wagner Coelho de Albuquerque Pereira3Cícero Ferreira Fernandes Costa Filho4Centro de Tecnologia Eletrônica e da Informação/Universidade Federal do AmazonasCentro de Tecnologia Eletrônica e da Informação/Universidade Federal do AmazonasCentro de Tecnologia Eletrônica e da Informação/Universidade Federal do AmazonasPrograma de Engenharia Biomédica/COPPE/Universidade Federal do Rio de JaneiroCentro de Tecnologia Eletrônica e da Informação/Universidade Federal do AmazonasAbstract Background Outlining lesion contours in Ultra Sound (US) breast images is an important step in breast cancer diagnosis. Malignant lesions infiltrate the surrounding tissue, generating irregular contours, with spiculation and angulated margins, whereas benign lesions produce contours with a smooth outline and elliptical shape. In breast imaging, the majority of the existing publications in the literature focus on using Convolutional Neural Networks (CNNs) for segmentation and classification of lesions in mammographic images. In this study our main objective is to assess the ability of CNNs in detecting contour irregularities in breast lesions in US images. Methods In this study we compare the performance of two CNNs with Direct Acyclic Graph (DAG) architecture and one CNN with a series architecture for breast lesion segmentation in US images. DAG and series architectures are both feedforward networks. The difference is that a DAG architecture could have more than one path between the first layer and end layer, whereas a series architecture has only one path from the beginning layer to the end layer. The CNN architectures were evaluated with two datasets. Results With the more complex DAG architecture, the following mean values were obtained for the metrics used to evaluate the segmented contours: global accuracy: 0.956; IOU: 0.876; F measure: 68.77%; Dice coefficient: 0.892. Conclusion The CNN DAG architecture shows the best metric values used for quantitatively evaluating the segmented contours compared with the gold-standard contours. The segmented contours obtained with this architecture also have more details and irregularities, like the gold-standard contours.http://link.springer.com/article/10.1186/s12880-019-0389-2Breast lesionUltrasound imagesConvolutional neural networks |
| spellingShingle | Marly Guimarães Fernandes Costa João Paulo Mendes Campos Gustavo de Aquino e Aquino Wagner Coelho de Albuquerque Pereira Cícero Ferreira Fernandes Costa Filho Evaluating the performance of convolutional neural networks with direct acyclic graph architectures in automatic segmentation of breast lesion in US images BMC Medical Imaging Breast lesion Ultrasound images Convolutional neural networks |
| title | Evaluating the performance of convolutional neural networks with direct acyclic graph architectures in automatic segmentation of breast lesion in US images |
| title_full | Evaluating the performance of convolutional neural networks with direct acyclic graph architectures in automatic segmentation of breast lesion in US images |
| title_fullStr | Evaluating the performance of convolutional neural networks with direct acyclic graph architectures in automatic segmentation of breast lesion in US images |
| title_full_unstemmed | Evaluating the performance of convolutional neural networks with direct acyclic graph architectures in automatic segmentation of breast lesion in US images |
| title_short | Evaluating the performance of convolutional neural networks with direct acyclic graph architectures in automatic segmentation of breast lesion in US images |
| title_sort | evaluating the performance of convolutional neural networks with direct acyclic graph architectures in automatic segmentation of breast lesion in us images |
| topic | Breast lesion Ultrasound images Convolutional neural networks |
| url | http://link.springer.com/article/10.1186/s12880-019-0389-2 |
| work_keys_str_mv | AT marlyguimaraesfernandescosta evaluatingtheperformanceofconvolutionalneuralnetworkswithdirectacyclicgrapharchitecturesinautomaticsegmentationofbreastlesioninusimages AT joaopaulomendescampos evaluatingtheperformanceofconvolutionalneuralnetworkswithdirectacyclicgrapharchitecturesinautomaticsegmentationofbreastlesioninusimages AT gustavodeaquinoeaquino evaluatingtheperformanceofconvolutionalneuralnetworkswithdirectacyclicgrapharchitecturesinautomaticsegmentationofbreastlesioninusimages AT wagnercoelhodealbuquerquepereira evaluatingtheperformanceofconvolutionalneuralnetworkswithdirectacyclicgrapharchitecturesinautomaticsegmentationofbreastlesioninusimages AT ciceroferreirafernandescostafilho evaluatingtheperformanceofconvolutionalneuralnetworkswithdirectacyclicgrapharchitecturesinautomaticsegmentationofbreastlesioninusimages |