Thesis
A Deep Learning Approach to Detecting Dysphagia in Videofluoroscopy
University of Iowa
Master of Science (MS), University of Iowa
Spring 2020
DOI: 10.17077/etd.005452
Abstract
Dysphagia, or a disorder classified by difficulty swallowing, is a severe health problem that reduces the quality of life of those affected. The standard method to diagnose dysphagia is the x-ray video fluoroscopic swallowing exam (VFSE). In this paper, we investigate the use of deep learning networks to classify VFSE as normal or abnormal. We have based our network on a long term recurrent convolutional network (LRCN). 1154 VFSE were available to train the network. Using 10-fold cross-validation, the accuracy of classification was 85%, and the area under the ROC curve was 0.89. This work shows the promise of using deep learning networks as a screening tool to detect dysphagia in VFSE.
Details
- Title: Subtitle
- A Deep Learning Approach to Detecting Dysphagia in Videofluoroscopy
- Creators
- Patrick T. Wilhelm
- Contributors
- Joseph M Reinhardt (Advisor)Hans Johnson (Committee Member)Douglas J Van Daele (Committee Member)
- Resource Type
- Thesis
- Degree Awarded
- Master of Science (MS), University of Iowa
- Degree in
- Electrical and Computer Engineering
- Date degree season
- Spring 2020
- DOI
- 10.17077/etd.005452
- Publisher
- University of Iowa
- Number of pages
- ix, 42 pages
- Copyright
- Copyright 2020 Patrick T. Wilhelm
- Language
- English
- Description illustrations
- illustrations (some color)
- Description bibliographic
- Includes bibliographical references (pages 36-40).
- Public Abstract (ETD)
People are always looking for ways to improve or simplify their lives. Physicians are no exception. However, in the world of medicine, diagnosis is typically only possible through considerable time and effort on the part of clinicians. We set out to create a system that relieves some of the burden by automatically evaluating swallowing exam videos for the presence of a disease.
The disease, in this case, is dysphagia, or the medical term for swallowing difficulties. The basis of this system is a method that attempts to take features from a swallowing exam and see how they change over time. As the system examines these videos, the accuracy should increase. This learning is a feature of the system, which mimics how we, as humans, learn. We attempt to perform something, are given feedback about what we’ve done right or wrong, and the next time we experience that situation, we (hopefully) perform better.
Ultimately, we were able to create a system that detects dysphagia in a video and does it with reasonable accuracy. Though not perfect, it does provide a basis for future works to create screening tools that should help with the workloads of our doctors.
- Academic Unit
- Electrical and Computer Engineering
- Record Identifier
- 9983956192102771
Metrics
83 File views/ downloads
122 Record Views