Ren, Xueying
(2022)
Identifying the Spatial Scales of Neural Representations in Human Visual System using Functional MRI with Wavelet Transforms.
Master's Thesis, University of Pittsburgh.
(Unpublished)
Abstract
Visual information is thought to be processed in a hierarchical fashion from simple features processing in early visual areas to more complex information processing in higher-order visual areas. However, much remains to be learned about how the human visual system represents information, and whether information is represented at fine or coarse spatial scales. Traditional methods, such as spatial smoothing, which have been used to investigate spatial scales, cannot confine individual spatial scales stringently. In this study, we applied a novel method– the wavelet transforms – that could overcome these limitations in quantifying spatial scales. The goals of our study were two-fold: we will examine 1) if information is represented at fine or coarse spatial scales in various sub-regions in human visual system, and 2) if there is any correlation between the spatial-scale dependent information representations and receptive field size across different sub-regions in human visual system. To answer those questions, we applied a dual-tree complex wavelet transform (dt-CWT) to fMRI data volumes, which were collected as eight participants viewed four visual categories. Five orthogonal spatial scales were generated using the wavelets, and a set of new features were defined that took scale and directionality information into account for information representations. Those new features generated by the wavelets were then submitted to a multi-class classification analysis with a XGBoost machine learning algorithm. We adopted a Bayesian approach to evaluate the statistical significance of the classification results. Contradicting with earlier findings, we did not find evidence to support multi-scale dependent neural representations in human visual system. However, our results should be interpreted with caution as we looked at neural representations from a different perspective. Further studies are needed to gain more insights into the multi-scale dependent neural representations.
Share
| Citation/Export: |
|
| Social Networking: |
|
Details
| Item Type: |
University of Pittsburgh ETD
|
| Status: |
Unpublished |
| Creators/Authors: |
|
| ETD Committee: |
|
| Date: |
6 June 2022 |
| Date Type: |
Publication |
| Defense Date: |
1 July 2021 |
| Approval Date: |
6 June 2022 |
| Submission Date: |
17 August 2021 |
| Access Restriction: |
No restriction; Release the ETD for access worldwide immediately. |
| Number of Pages: |
35 |
| Institution: |
University of Pittsburgh |
| Schools and Programs: |
Dietrich School of Arts and Sciences > Psychology |
| Degree: |
MS - Master of Science |
| Thesis Type: |
Master's Thesis |
| Refereed: |
Yes |
| Uncontrolled Keywords: |
spatial scales, wavelets, visual system, fMRI |
| Date Deposited: |
06 Jun 2022 15:53 |
| Last Modified: |
06 Jun 2022 15:53 |
| URI: |
http://d-scholarship.pitt.edu/id/eprint/41682 |
Metrics
Monthly Views for the past 3 years
Plum Analytics
Actions (login required)
 |
View Item |
![[feed]](/images/feed-icon-32x32.png){kind=icon}