DermoNet: densely linked convolutional neural network for efficient skin lesion segmentation
Abstract Recent state-of-the-art methods for skin lesion segmentation are based on convolutional neural networks (CNNs). Even though these CNN-based segmentation approaches are accurate, they are computationally expensive. In this paper, we address this problem and propose an efficient fully convolu...
| Main Authors: | , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
SpringerOpen
2019-07-01
|
| Series: | EURASIP Journal on Image and Video Processing |
| Subjects: | |
| Online Access: | http://link.springer.com/article/10.1186/s13640-019-0467-y |
| _version_ | 1818215918408302592 |
|---|---|
| author | Saleh Baghersalimi Behzad Bozorgtabar Philippe Schmid-Saugeon Hazım Kemal Ekenel Jean-Philippe Thiran |
| author_facet | Saleh Baghersalimi Behzad Bozorgtabar Philippe Schmid-Saugeon Hazım Kemal Ekenel Jean-Philippe Thiran |
| author_sort | Saleh Baghersalimi |
| collection | DOAJ |
| description | Abstract Recent state-of-the-art methods for skin lesion segmentation are based on convolutional neural networks (CNNs). Even though these CNN-based segmentation approaches are accurate, they are computationally expensive. In this paper, we address this problem and propose an efficient fully convolutional neural network, named DermoNet. In DermoNet, due to our densely connected convolutional blocks and skip connections, network layers can reuse information from their preceding layers and ensure high accuracy in later network layers. By doing so, we take advantage of the capability of high-level feature representations learned at intermediate layers with varying scales and resolutions for lesion segmentation. Quantitative evaluation is conducted on three well-established public benchmark datasets: the ISBI 2016, ISBI 2017, and the PH2 datasets. The experimental results show that our proposed approach outperforms the state-of-the-art algorithms on these three datasets. We also compared the runtime performance of DermoNet with two other related architectures, which are fully convolutional networks and U-Net. The proposed approach is found to be faster and suitable for practical applications. |
| first_indexed | 2024-12-12T06:43:43Z |
| format | Article |
| id | doaj.art-cb72aff5a8df4c5f935c53df53a533de |
| institution | Directory Open Access Journal |
| issn | 1687-5281 |
| language | English |
| last_indexed | 2024-12-12T06:43:43Z |
| publishDate | 2019-07-01 |
| publisher | SpringerOpen |
| record_format | Article |
| series | EURASIP Journal on Image and Video Processing |
| spelling | doaj.art-cb72aff5a8df4c5f935c53df53a533de2022-12-22T00:34:16ZengSpringerOpenEURASIP Journal on Image and Video Processing1687-52812019-07-012019111010.1186/s13640-019-0467-yDermoNet: densely linked convolutional neural network for efficient skin lesion segmentationSaleh Baghersalimi0Behzad Bozorgtabar1Philippe Schmid-Saugeon2Hazım Kemal Ekenel3Jean-Philippe Thiran4Electrical Engineering Department, Signal Processing Laboratory (LTS5), École Polytechnique Fédérale de Lausanne (EPFL)Electrical Engineering Department, Signal Processing Laboratory (LTS5), École Polytechnique Fédérale de Lausanne (EPFL)DermoSafe SA, EPFL Innovation ParkDepartment of Computer EngineeringElectrical Engineering Department, Signal Processing Laboratory (LTS5), École Polytechnique Fédérale de Lausanne (EPFL)Abstract Recent state-of-the-art methods for skin lesion segmentation are based on convolutional neural networks (CNNs). Even though these CNN-based segmentation approaches are accurate, they are computationally expensive. In this paper, we address this problem and propose an efficient fully convolutional neural network, named DermoNet. In DermoNet, due to our densely connected convolutional blocks and skip connections, network layers can reuse information from their preceding layers and ensure high accuracy in later network layers. By doing so, we take advantage of the capability of high-level feature representations learned at intermediate layers with varying scales and resolutions for lesion segmentation. Quantitative evaluation is conducted on three well-established public benchmark datasets: the ISBI 2016, ISBI 2017, and the PH2 datasets. The experimental results show that our proposed approach outperforms the state-of-the-art algorithms on these three datasets. We also compared the runtime performance of DermoNet with two other related architectures, which are fully convolutional networks and U-Net. The proposed approach is found to be faster and suitable for practical applications.http://link.springer.com/article/10.1186/s13640-019-0467-yFully convolutional neural networksLesion segmentation |
| spellingShingle | Saleh Baghersalimi Behzad Bozorgtabar Philippe Schmid-Saugeon Hazım Kemal Ekenel Jean-Philippe Thiran DermoNet: densely linked convolutional neural network for efficient skin lesion segmentation EURASIP Journal on Image and Video Processing Fully convolutional neural networks Lesion segmentation |
| title | DermoNet: densely linked convolutional neural network for efficient skin lesion segmentation |
| title_full | DermoNet: densely linked convolutional neural network for efficient skin lesion segmentation |
| title_fullStr | DermoNet: densely linked convolutional neural network for efficient skin lesion segmentation |
| title_full_unstemmed | DermoNet: densely linked convolutional neural network for efficient skin lesion segmentation |
| title_short | DermoNet: densely linked convolutional neural network for efficient skin lesion segmentation |
| title_sort | dermonet densely linked convolutional neural network for efficient skin lesion segmentation |
| topic | Fully convolutional neural networks Lesion segmentation |
| url | http://link.springer.com/article/10.1186/s13640-019-0467-y |
| work_keys_str_mv | AT salehbaghersalimi dermonetdenselylinkedconvolutionalneuralnetworkforefficientskinlesionsegmentation AT behzadbozorgtabar dermonetdenselylinkedconvolutionalneuralnetworkforefficientskinlesionsegmentation AT philippeschmidsaugeon dermonetdenselylinkedconvolutionalneuralnetworkforefficientskinlesionsegmentation AT hazımkemalekenel dermonetdenselylinkedconvolutionalneuralnetworkforefficientskinlesionsegmentation AT jeanphilippethiran dermonetdenselylinkedconvolutionalneuralnetworkforefficientskinlesionsegmentation |