Link to the University of Pittsburgh Homepage
Link to the University Library System Homepage Link to the Contact Us Form

Independent Components Analysis of Local Field Potentials Sources in the Primate Superior Colliculus

LIU, YU (2019) Independent Components Analysis of Local Field Potentials Sources in the Primate Superior Colliculus. Master's Thesis, University of Pittsburgh. (Unpublished)

Download (4MB) | Preview


When visually-guided eye movements (saccades) are produced, some signal-available electrical potentials are recorded by electrodes in the superior colliculus (SC), which is ideal to investigate communication between the different layers due to the laminar organization and structured anatomical connections within SC. The first step to realize this objective is identifying the location of signal-generators involved in the procedure. Local Field Potential (LFP), representing a combination of the various generators measured in the vicinity of each electrodes, makes this issue as similar as blind source separation problem. Thus, Independent Component Analysis (ICA) is proposed to analyze the specific spatial and temporal properties of LFP recorded in Rhesus monkeys performing a standard delayed saccade task. After being preprocessed, the LFP data is decomposed into two parts: mixing matrix and independent component (IC). The LFPs are binned into segments, and the ICs and their corresponding mixing weights are computed for each epoch.

We also compared our results against current source density (CSD) techniques, which relies on the second spatial derivative of the mixing coefficients computed with a Laplacian of Gaussian kernel. We then reconstructed the CSD that corresponds to each component by multiplying that IC by the computed derivative of its coefficients. To identify the dominant ICs, we performed correlation analysis between each CSD reconstructed via a single IC against the ground truth CSD (the second spatial derivative of the raw LFP) and we kept only those ICs with a correlation score exceeding a predefined threshold. These dominant ICs correspond to the current generators in this time window, and the spatial distribution of the coefficients defines the depth at which this sink originates. Our analysis indicates that the correlation between the IC based reconstruction and the ground truth CSD exceeds 90% in best. We consistently identified 2-3 generator sites, mostly confined near the boundary of superficial and intermediate layers. The strongest modulations occurred slightly after saccade onset and in response to target onset. Thus, the ICA method can accurately reconstruct the CSD, as well as identify and quantify the origin of current sinks.


Social Networking:
Share |


Item Type: University of Pittsburgh ETD
Status: Unpublished
CreatorsEmailPitt UsernameORCID
LIU, YUyul189@pitt.eduyul189
ETD Committee:
TitleMemberEmail AddressPitt UsernameORCID
Thesis AdvisorDallal,
Thesis AdvisorGandhi,
Committee MemberEl-Jaroudi, Amro
Committee MemberMao, Zhi-Hong
Date: 10 September 2019
Date Type: Publication
Defense Date: 17 July 2019
Approval Date: 10 September 2019
Submission Date: 17 July 2019
Access Restriction: No restriction; Release the ETD for access worldwide immediately.
Number of Pages: 77
Institution: University of Pittsburgh
Schools and Programs: Swanson School of Engineering > Electrical Engineering
Degree: MS - Master of Science
Thesis Type: Master's Thesis
Refereed: Yes
Uncontrolled Keywords: ICA, LFP, CSD, superior colliculus
Date Deposited: 10 Sep 2019 14:15
Last Modified: 10 Sep 2019 14:15


Monthly Views for the past 3 years

Plum Analytics

Actions (login required)

View Item View Item