Automatic detection, segmentation and classification of abdominal aortic aneurysm using deep learning
This study is focused on developing an automated algorithm for the detection and segmentation of Abdominal Aortic Aneurysm (AAA) region in CT Angiography images. The outcome of this research will offer great assistance for radiologists to detect the AAA region and estimate its volume in CT datasets...
| Main Author: | |
|---|---|
| Format: | Thesis |
| Language: | English |
| Published: |
2015
|
| Subjects: | |
| Online Access: | http://eprints.utm.my/81027/1/HoAikHongMFKE2015.pdf |
| _version_ | 1796863306158309376 |
|---|---|
| author | Ho, Aik Hong |
| author_facet | Ho, Aik Hong |
| author_sort | Ho, Aik Hong |
| collection | ePrints |
| description | This study is focused on developing an automated algorithm for the detection and segmentation of Abdominal Aortic Aneurysm (AAA) region in CT Angiography images. The outcome of this research will offer great assistance for radiologists to detect the AAA region and estimate its volume in CT datasets efficiently. In addition, suitable treatment strategies can also be suggested based on the classification of the AAA severity and measurement of the aorta diameter. This research takes the initiative by exploring and applying deep learning architecture in the study of AAA detection and segmentation, which has never been done by other researchers before in AAA problems. The findings from this study will also act as a reference for other similar future works. Deep Belief Network (DBN) is applied for the purpose of AAA detection and severity classification in this study. Optimum parameters for training the DBN are determined for the training data from the selected dataset. AAA region can be successfully segmented from the CT images and the result is comparable to the existing method with advantage over the existing method that the proposed method is fully automatic and added with auto detection and classification features. The limitation of the trained DBN in AAA detection accuracy can be improved by incorporating and adjusting the detection probability threshold and shape constraints. In future, the DBN can be further enhanced by adding and training it with more data which covers a wider variety of features, as well as extending its capability to perform detailed segmentation on AAA region. |
| first_indexed | 2024-03-05T20:24:46Z |
| format | Thesis |
| id | utm.eprints-81027 |
| institution | Universiti Teknologi Malaysia - ePrints |
| language | English |
| last_indexed | 2024-03-05T20:24:46Z |
| publishDate | 2015 |
| record_format | dspace |
| spelling | utm.eprints-810272019-07-24T03:06:05Z http://eprints.utm.my/81027/ Automatic detection, segmentation and classification of abdominal aortic aneurysm using deep learning Ho, Aik Hong TK Electrical engineering. Electronics Nuclear engineering This study is focused on developing an automated algorithm for the detection and segmentation of Abdominal Aortic Aneurysm (AAA) region in CT Angiography images. The outcome of this research will offer great assistance for radiologists to detect the AAA region and estimate its volume in CT datasets efficiently. In addition, suitable treatment strategies can also be suggested based on the classification of the AAA severity and measurement of the aorta diameter. This research takes the initiative by exploring and applying deep learning architecture in the study of AAA detection and segmentation, which has never been done by other researchers before in AAA problems. The findings from this study will also act as a reference for other similar future works. Deep Belief Network (DBN) is applied for the purpose of AAA detection and severity classification in this study. Optimum parameters for training the DBN are determined for the training data from the selected dataset. AAA region can be successfully segmented from the CT images and the result is comparable to the existing method with advantage over the existing method that the proposed method is fully automatic and added with auto detection and classification features. The limitation of the trained DBN in AAA detection accuracy can be improved by incorporating and adjusting the detection probability threshold and shape constraints. In future, the DBN can be further enhanced by adding and training it with more data which covers a wider variety of features, as well as extending its capability to perform detailed segmentation on AAA region. 2015-06 Thesis NonPeerReviewed application/pdf en http://eprints.utm.my/81027/1/HoAikHongMFKE2015.pdf Ho, Aik Hong (2015) Automatic detection, segmentation and classification of abdominal aortic aneurysm using deep learning. Masters thesis, Universiti Teknologi Malaysia, Faculty of Electrical Engineering. http://dms.library.utm.my:8080/vital/access/manager/Repository/vital:121241 |
| spellingShingle | TK Electrical engineering. Electronics Nuclear engineering Ho, Aik Hong Automatic detection, segmentation and classification of abdominal aortic aneurysm using deep learning |
| title | Automatic detection, segmentation and classification of abdominal aortic aneurysm using deep learning |
| title_full | Automatic detection, segmentation and classification of abdominal aortic aneurysm using deep learning |
| title_fullStr | Automatic detection, segmentation and classification of abdominal aortic aneurysm using deep learning |
| title_full_unstemmed | Automatic detection, segmentation and classification of abdominal aortic aneurysm using deep learning |
| title_short | Automatic detection, segmentation and classification of abdominal aortic aneurysm using deep learning |
| title_sort | automatic detection segmentation and classification of abdominal aortic aneurysm using deep learning |
| topic | TK Electrical engineering. Electronics Nuclear engineering |
| url | http://eprints.utm.my/81027/1/HoAikHongMFKE2015.pdf |
| work_keys_str_mv | AT hoaikhong automaticdetectionsegmentationandclassificationofabdominalaorticaneurysmusingdeeplearning |