Wu, Yitao
(2021)
A Novel Deep Learning Framework to Identify Latent Neuroendophenotypes from Multimodal Brain Imaging Data.
Master's Thesis, University of Pittsburgh.
(Unpublished)
Abstract
The expertise required to ensure adequate treatment for patients with complex cases is significantly deficient, which leads to the high demand for subtyping or clustering analysis on different clinical situations. The identification and refinement of disease-related subtypes will support both medical treatments and pathological research. Clinically, clustering can narrow down the possible causes and provide effective treatment options. However, the clustering on non-invasive multimodal brain imaging data has not been well addressed.
In this thesis, we explore this clustering issue using a deep unsupervised embedded clustering (DEMC) method on multimodal brain imaging data. T1-weighted magnetic resonance imaging (MRI) features and resting-state functional MRI-derived brain networks are learned by a sparse autoencoder and a stacked autoencoder separately and then transformed into the embedding space. Then, the K-Means approach was adopted to set the initial center of the deeply embedded clustering structure (DEC) as the centroids, after which DEC clusters with the KL divergence. In the entire processing, the deep embedding and clustering are optimized simultaneously. This new framework was tested on 994 subjects from Human Connectome Project (HCP) and the results show that this new framework has better clustering performance in comparison with other benchmark algorithms.
Share
| Citation/Export: |
|
| Social Networking: |
|
Details
| Item Type: |
University of Pittsburgh ETD
|
| Status: |
Unpublished |
| Creators/Authors: |
|
| ETD Committee: |
|
| Date: |
13 June 2021 |
| Date Type: |
Publication |
| Defense Date: |
8 April 2021 |
| Approval Date: |
13 June 2021 |
| Submission Date: |
8 April 2021 |
| Access Restriction: |
No restriction; Release the ETD for access worldwide immediately. |
| Number of Pages: |
37 |
| Institution: |
University of Pittsburgh |
| Schools and Programs: |
Swanson School of Engineering > Electrical and Computer Engineering |
| Degree: |
MS - Master of Science |
| Thesis Type: |
Master's Thesis |
| Refereed: |
Yes |
| Uncontrolled Keywords: |
Multimodal Clustering, Deep learning, Autoencoder, Deep Embedded Clustering, Brain imaging data |
| Date Deposited: |
13 Jun 2021 18:43 |
| Last Modified: |
13 Jun 2021 18:43 |
| URI: |
http://d-scholarship.pitt.edu/id/eprint/40543 |
Metrics
Monthly Views for the past 3 years
Plum Analytics
Actions (login required)
 |
View Item |
![[feed]](/images/feed-icon-32x32.png){kind=icon}