Full Text Available
Note: Clicking the button above will open the full text document at the original institutional repository in a new window.
Semi-Supervised Transfer Learning for medical images as an alternative to ImageNet Transfer Learning
One of the main disadvantages of supervised transfer learning is that it necessarily requires a large amount of expensive manually labelled training data. Consequently, even in medical imaging, transfer learning from natural image datasets (such as ImageNet) has become the norm. However, this approa...
Saved in:
| Main Author: | |
|---|---|
| Other Authors: | |
| Format: | Thesis |
| Language: | English |
| Published: |
Department of Electrical Engineering
2023
|
| Subjects: | |
| Tags: |
No Tags, Be the first to tag this record!
|
| _version_ | 1867614423336615936 |
|---|---|
| access_status_str | Open Access |
| author | Nkwentsha, Xolisani |
| author2 | Nicolls, Fred |
| author_browse | Nicolls, Fred Nkwentsha, Xolisani |
| author_facet | Nicolls, Fred Nkwentsha, Xolisani |
| author_sort | Nkwentsha, Xolisani |
| collection | Thesis |
| description | One of the main disadvantages of supervised transfer learning is that it necessarily requires a large amount of expensive manually labelled training data. Consequently, even in medical imaging, transfer learning from natural image datasets (such as ImageNet) has become the norm. However, this approach has been shown to be ineffective due to the significant differences between medical images and natural images. Developing a large-scale medical imaging dataset for transfer learning would be too expensive, therefore the possibility of using large amounts of unlabelled data for feature learning is very attractive. In this work, we propose a semi-supervised transfer learning method for training deep learning models for medical imaging. The main idea behind the proposed method is to leverage unlabelled medical image datasets to improve accuracy for the target task by transferring feature maps learned from an unsupervised task to the supervised target task. We leverage unlabelled data by transferring weights/kernels and representations learned by an autoencoder (specifically the encoder part) during a reconstruction task to a classification task. We show the applicability of features learned by the autoencoder from the collection of unlabelled x-ray images to a pneumonia classification problem. Our proposed method improves the baseline performance by 4.167% in accuracy and the precision, recall and F1 score by 4%. We also demonstrate that increasing the size of the unlabelled dataset used to train the autoencoder improves the performance on the target task. This increase in the size of the dataset resulted in an overall 5.288% accuracy increase from the baseline. We also compare our method with ImageNet models on the target dataset. For the standard ImageNet architectures, we evaluate ResNet50 and Inception-v3, which have both been used extensively in medical deep learning applications. Our proposed method outperforms both standard ImageNet models on the target task. These results demonstrate that learning features from unlabelled medical images for transfer learning for medical imaging tasks is more effective than transfer learning from natural images, at least for the problem of pneumonia detection. |
| format | Thesis |
| id | oai:open.uct.ac.za:11427/37687 |
| institution | University of Cape Town (South Africa) |
| language | eng |
| last_indexed | 2026-06-10T12:51:48.486Z |
| license_str | Not specified — see source repository |
| provenance_str_mv | Harvested via OAI-PMH from UCTD — University of Cape Town Open Access Repository |
| publishDate | 2023 |
| publishDateRange | 2023 |
| publishDateSort | 2023 |
| publisher | Department of Electrical Engineering |
| publisherStr | Department of Electrical Engineering |
| record_format | dspace |
| source_str | UCTD — University of Cape Town Open Access Repository |
| spelling | oai:open.uct.ac.za:11427/37687 Semi-Supervised Transfer Learning for medical images as an alternative to ImageNet Transfer Learning Nkwentsha, Xolisani Nicolls, Fred Electrical Engineering One of the main disadvantages of supervised transfer learning is that it necessarily requires a large amount of expensive manually labelled training data. Consequently, even in medical imaging, transfer learning from natural image datasets (such as ImageNet) has become the norm. However, this approach has been shown to be ineffective due to the significant differences between medical images and natural images. Developing a large-scale medical imaging dataset for transfer learning would be too expensive, therefore the possibility of using large amounts of unlabelled data for feature learning is very attractive. In this work, we propose a semi-supervised transfer learning method for training deep learning models for medical imaging. The main idea behind the proposed method is to leverage unlabelled medical image datasets to improve accuracy for the target task by transferring feature maps learned from an unsupervised task to the supervised target task. We leverage unlabelled data by transferring weights/kernels and representations learned by an autoencoder (specifically the encoder part) during a reconstruction task to a classification task. We show the applicability of features learned by the autoencoder from the collection of unlabelled x-ray images to a pneumonia classification problem. Our proposed method improves the baseline performance by 4.167% in accuracy and the precision, recall and F1 score by 4%. We also demonstrate that increasing the size of the unlabelled dataset used to train the autoencoder improves the performance on the target task. This increase in the size of the dataset resulted in an overall 5.288% accuracy increase from the baseline. We also compare our method with ImageNet models on the target dataset. For the standard ImageNet architectures, we evaluate ResNet50 and Inception-v3, which have both been used extensively in medical deep learning applications. Our proposed method outperforms both standard ImageNet models on the target task. These results demonstrate that learning features from unlabelled medical images for transfer learning for medical imaging tasks is more effective than transfer learning from natural images, at least for the problem of pneumonia detection. 2023-04-13T07:45:56Z 2023-04-13T07:45:56Z 2022 2023-04-12T11:02:49Z Master Thesis Masters MSc http://hdl.handle.net/11427/37687 eng application/pdf Department of Electrical Engineering Faculty of Engineering and the Built Environment |
| spellingShingle | Electrical Engineering Nkwentsha, Xolisani Semi-Supervised Transfer Learning for medical images as an alternative to ImageNet Transfer Learning |
| thesis_degree_str | Master's |
| title | Semi-Supervised Transfer Learning for medical images as an alternative to ImageNet Transfer Learning |
| title_full | Semi-Supervised Transfer Learning for medical images as an alternative to ImageNet Transfer Learning |
| title_fullStr | Semi-Supervised Transfer Learning for medical images as an alternative to ImageNet Transfer Learning |
| title_full_unstemmed | Semi-Supervised Transfer Learning for medical images as an alternative to ImageNet Transfer Learning |
| title_short | Semi-Supervised Transfer Learning for medical images as an alternative to ImageNet Transfer Learning |
| title_sort | semi supervised transfer learning for medical images as an alternative to imagenet transfer learning |
| topic | Electrical Engineering |
| url | http://hdl.handle.net/11427/37687 |
| work_keys_str_mv | AT nkwentshaxolisani semisupervisedtransferlearningformedicalimagesasanalternativetoimagenettransferlearning |