Unsupervised Learning for Concept Detection in Medical Images: A Comparative Analysis
As digital medical imaging becomes more prevalent and archives increase in size, representation learning exposes an interesting opportunity for enhanced medical decision support systems. On the other hand, medical imaging data is often scarce and short on annotations. In this paper, we present an as...
| Main Authors: | , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
MDPI AG
2018-07-01
|
| Series: | Applied Sciences |
| Subjects: | |
| Online Access: | http://www.mdpi.com/2076-3417/8/8/1213 |
| _version_ | 1819171179127111680 |
|---|---|
| author | Eduardo Pinho Carlos Costa |
| author_facet | Eduardo Pinho Carlos Costa |
| author_sort | Eduardo Pinho |
| collection | DOAJ |
| description | As digital medical imaging becomes more prevalent and archives increase in size, representation learning exposes an interesting opportunity for enhanced medical decision support systems. On the other hand, medical imaging data is often scarce and short on annotations. In this paper, we present an assessment of unsupervised feature learning approaches for images in biomedical literature which can be applied to automatic biomedical concept detection. Six unsupervised representation learning methods were built, including traditional bags of visual words, autoencoders, and generative adversarial networks. Each model was trained, and their respective feature spaces evaluated using images from the ImageCLEF 2017 concept detection task. The highest mean F1 score of 0.108 was obtained using representations from an adversarial autoencoder, which increased to 0.111 when combined with the representations from the sparse denoising autoencoder. We conclude that it is possible to obtain more powerful representations with modern deep learning approaches than with previously popular computer vision methods. The possibility of semi-supervised learning as well as its use in medical information retrieval problems are the next steps to be strongly considered. |
| first_indexed | 2024-12-22T19:47:10Z |
| format | Article |
| id | doaj.art-7cae67042f594b2cb53d678521d1901f |
| institution | Directory Open Access Journal |
| issn | 2076-3417 |
| language | English |
| last_indexed | 2024-12-22T19:47:10Z |
| publishDate | 2018-07-01 |
| publisher | MDPI AG |
| record_format | Article |
| series | Applied Sciences |
| spelling | doaj.art-7cae67042f594b2cb53d678521d1901f2022-12-21T18:14:40ZengMDPI AGApplied Sciences2076-34172018-07-0188121310.3390/app8081213app8081213Unsupervised Learning for Concept Detection in Medical Images: A Comparative AnalysisEduardo Pinho0Carlos Costa1Institute of Electronics and Informatics Engineering of Aveiro, University of Aveiro; Campus Universitário de Santiago, 3810-193 Aveiro, PortugalInstitute of Electronics and Informatics Engineering of Aveiro, University of Aveiro; Campus Universitário de Santiago, 3810-193 Aveiro, PortugalAs digital medical imaging becomes more prevalent and archives increase in size, representation learning exposes an interesting opportunity for enhanced medical decision support systems. On the other hand, medical imaging data is often scarce and short on annotations. In this paper, we present an assessment of unsupervised feature learning approaches for images in biomedical literature which can be applied to automatic biomedical concept detection. Six unsupervised representation learning methods were built, including traditional bags of visual words, autoencoders, and generative adversarial networks. Each model was trained, and their respective feature spaces evaluated using images from the ImageCLEF 2017 concept detection task. The highest mean F1 score of 0.108 was obtained using representations from an adversarial autoencoder, which increased to 0.111 when combined with the representations from the sparse denoising autoencoder. We conclude that it is possible to obtain more powerful representations with modern deep learning approaches than with previously popular computer vision methods. The possibility of semi-supervised learning as well as its use in medical information retrieval problems are the next steps to be strongly considered.http://www.mdpi.com/2076-3417/8/8/1213representation learningunsupervised learningdeep learningcontent-based image retrieval |
| spellingShingle | Eduardo Pinho Carlos Costa Unsupervised Learning for Concept Detection in Medical Images: A Comparative Analysis Applied Sciences representation learning unsupervised learning deep learning content-based image retrieval |
| title | Unsupervised Learning for Concept Detection in Medical Images: A Comparative Analysis |
| title_full | Unsupervised Learning for Concept Detection in Medical Images: A Comparative Analysis |
| title_fullStr | Unsupervised Learning for Concept Detection in Medical Images: A Comparative Analysis |
| title_full_unstemmed | Unsupervised Learning for Concept Detection in Medical Images: A Comparative Analysis |
| title_short | Unsupervised Learning for Concept Detection in Medical Images: A Comparative Analysis |
| title_sort | unsupervised learning for concept detection in medical images a comparative analysis |
| topic | representation learning unsupervised learning deep learning content-based image retrieval |
| url | http://www.mdpi.com/2076-3417/8/8/1213 |
| work_keys_str_mv | AT eduardopinho unsupervisedlearningforconceptdetectioninmedicalimagesacomparativeanalysis AT carloscosta unsupervisedlearningforconceptdetectioninmedicalimagesacomparativeanalysis |