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

Development of Statistical Models for Functional Near-infrared Spectroscopy Data Analysis Incorporating Anatomical and Probe Registration Prior Information

Zhai, Xuetong (2021) Development of Statistical Models for Functional Near-infrared Spectroscopy Data Analysis Incorporating Anatomical and Probe Registration Prior Information. Doctoral Dissertation, University of Pittsburgh. (Unpublished)

Download (12MB) | Preview


Functional near-infrared spectroscopy (fNIRS) is a non-invasive technology that uses low-levels of non-ionizing light in the range of 650 -- 900 nm (red and near-infrared) to record changes in the optical absorption and scattering of tissue. In particular, oxy-hemoglobin (HbO) and deoxy-hemoglobin (HbR) have characteristic absorption spectra at these wavelengths, which are used to discriminate blood flow and oxygen metabolism changes. As compared with functional magnetic resonance imaging (fMRI), fNIRS is less costly, more portable, and allows for a wider range of experimental scenarios because it neither requires a dedicated scanner nor needs the subject to lay supine.

Current challenges in fNIRS data analysis include: (i) a small change in brain anatomy or optical probe positioning can create huge differences in fNIRS measurements even though the underlying brain activity remains the same due to the existence of ``blind-spots"; (ii) fNIRS image reconstruction is a high-dimensional, under-determined, and ill-posed problem, in which there are thousands of parameters to estimate while only tens of measurements available and existing methods notably overestimate the false positive rate; (iii) brain anatomical information has rarely been used in current fNIRS data analyses.

This dissertation proposes two new methods aiming to improve fNIRS data analysis and overcome these challenges -- one of which is a channel-space method based on anatomically defined region-of-interest (ROI) and the other one is an image reconstruction method incorporating anatomical and physiological prior information. The two methods are developed using advanced statistical models including a combination of regularization models and Bayesian hierarchical modeling. The performance of the two methods is validated via numerical simulations and evaluated using receiver operating characteristics (ROC)-based tools. The statistical comparisons with conventional methods suggest significant improvements.


Social Networking:
Share |


Item Type: University of Pittsburgh ETD
Status: Unpublished
CreatorsEmailPitt UsernameORCID
Zhai, Xuetongxuz19@pitt.eduxuz190000-0002-2618-5686
ETD Committee:
TitleMemberEmail AddressPitt UsernameORCID
Committee ChairHuppert, Theodorehuppert1@pitt.eduhuppert1
Committee MemberStetten, Georgestetten@pitt.edustetten
Committee MemberAizenstein, Howardaizensteinhj@upmc.eduaizen
Committee MemberKrafty,
Date: 26 January 2021
Date Type: Publication
Defense Date: 6 November 2020
Approval Date: 26 January 2021
Submission Date: 3 November 2020
Access Restriction: No restriction; Release the ETD for access worldwide immediately.
Number of Pages: 138
Institution: University of Pittsburgh
Schools and Programs: Swanson School of Engineering > Bioengineering
Degree: PhD - Doctor of Philosophy
Thesis Type: Doctoral Dissertation
Refereed: Yes
Uncontrolled Keywords: functional near-infrared spectroscopy; statistical analysis; functional brain imaging image reconstruction
Date Deposited: 26 Jan 2021 20:57
Last Modified: 26 Jan 2021 20:57


Monthly Views for the past 3 years

Plum Analytics

Actions (login required)

View Item View Item