Dissertation
Deep-learning-based differentiation between the causes of optic disc swelling using OCT and fundus photographs
University of Iowa
Doctor of Philosophy (PhD), University of Iowa
Spring 2021
DOI: 10.17077/etd.006033
Abstract
Swelling of the optic disc can be caused due to several underlying reasons among which some are life-threatening while some are benign. The correct determination of the cause of swelling may help the clinicians start an appropriate treatment process, reduce patient's anxiety, and ultimately, in extreme cases, save lives in emergency situations.
This doctoral work involves several deep-learning-based approaches to determine the cause of swelling, and also in the process, to estimate the severity of swelling from commonly available non-invasive imaging modalities. In order to determine the cause of swelling, important clinical features are required to be extracted, including features from the retinal blood vessels, retinal folds and peripapillary wrinkles.
The retinal blood vessels are an important feature to determine the cause and severity of swelling, for automated registration of fundus and OCT images and as a pre-processing step for further retinal structure analysis. Retinal folds and peripapillary wrinkles are also important features to determine the severity and cause of swelling. Hence, in this doctoral work, automated deep-learning-based approaches have been developed to segment these vessels and folds/wrinkles.
Since the regional, total retinal volumes and retinal shape measures have been useful in recent work to determine the cause and severity of swelling, this doctoral work also focuses on the development of a deep-learning-based approach which provides the total-retinal-thickness maps from fundus photographs. Thus fundus photographs can provide an estimation of regional and total swelling severity. The results are also used to estimate shape of the 3D optic-nerve-head from color fundus photographs.
Ultimately, the features have been used to differentiate between the causes of swelling using deep learning. The deep-learning approaches' performance has been compared with traditional-machine-learning-based approaches that use the retinal blood vessels, folds/wrinkles, regional swelling, and shape features extracted before. The deep-learning-based approaches reduce the need for custom-designed features and enable easier adaptation to new dataset.
Details
- Title: Subtitle
- Deep-learning-based differentiation between the causes of optic disc swelling using OCT and fundus photographs
- Creators
- Mohammad Shafkat Islam
- Contributors
- Mona K Garvin (Advisor)Stephen Baek (Committee Member)Tyler Bell (Committee Member)Matthews Jacob (Committee Member)Xiaodong Wu (Committee Member)
- Resource Type
- Dissertation
- Degree Awarded
- Doctor of Philosophy (PhD), University of Iowa
- Degree in
- Electrical and Computer Engineering
- Date degree season
- Spring 2021
- DOI
- 10.17077/etd.006033
- Publisher
- University of Iowa
- Number of pages
- xvi, 110 pages
- Copyright
- Copyright 2021 Mohammad Shafkat Islam
- Language
- English
- Description illustrations
- illustrations (some color)
- Description bibliographic
- Includes bibliographical references (pages 102-110).
- Public Abstract (ETD)
The optic nerve is located towards the back of the eye which connects the eye to the brain. The region which connects optic nerve to the eye is known as optic nerve head (ONH). This optic nerve head is visible without any invasive procedure, and hence, it can be used to detect any abnormal condition inside the brain. For example, if the optic nerve head appears swollen, it may indicate a number of serious underlying conditions. Some of these conditions may require immediate medical attention. In some cases however, the optic nerve head appears swollen, but actually, the reason is congenital.
Current approaches for detecting the reason of optic disc swelling are either based on invasive or non-invasive procedures. Most of this doctoral work focuses on development of automated systems to detect important clinical features and differentiate between the causes of optic disc swelling. The goal of the doctoral work is to assist the clinicians in emergency situations, using the readily available imaging modalities and provide prior knowledge about the clinical condition.
- Academic Unit
- Electrical and Computer Engineering
- Record Identifier
- 9984097077602771
Metrics
4 File views/ downloads
55 Record Views