Bilal Khan

Graduation Semester and Year

2009

Language

English

Document Type

Thesis

Degree Name

Master of Science in Biomedical Engineering

Department

Bioengineering

First Advisor

Georgios Alexandrakis

Abstract

Functional near infrared spectroscopy (fNIRS) is widely used to monitor hemodynamic changes occurring in the cerebral cortex as a result of neuronal activation. One of the current challenges in this field is that activation-related hemodynamic signals are often eclipsed by global hemodynamic fluctuations due to cardiac pulsation, respiration and Mayer waves. In this study, we demonstrate by using a combination of principal component analysis (PCA) and adaptive filtering that the global hemodynamic modulation signals can be effectively removed from the fNIRS measuring cortical hemodynamic response. While pediatric subjects were performing a finger tapping task, we concurrently recorded cardiac pulsation and respiration using a pulse oximeter and a piezo-electric transducer, respectively. The results indicate that fNIRS data quality is significantly improved, as quantified by temporal signal-to-noise ratio (SNR), contrast-to-noise ratio (CNR), and image SNR metrics after reducing the effects of physiological artifacts. The significant improvement in SNR (p < 0.0001), CNR (p < 0.0001), and image SNR (p = 0.0014) leads to the conclusion that concurrent hemodynamic, cardiac pulsation, and respiration signal acquisition and filtering methods may need to become standard procedure in fNIRS, neuroimaging protocols. The filtered data can potentially be used to differentiate between motor cortex activation patterns in normal children and ones affected by cerebral palsy. The ratios of duration over time-to-peak temporal metrics were found to be significantly different (p < 0.0001) between normal subjects (1.28 +/- 0.23) and cerebral palsy subjects tapping with their affected hand (0.68 +/- 0.09). In the time-averaged reconstructed images, the distance of activation areas from the middle of the motor cortex in the ipsilateral hemisphere of the tapping hand were also found to be significantly different (p < 0.0001) between normal subjects (7.70 cm +/- 1.34 cm) and cerebral palsy subjects (3.50 cm +/- 1.78 cm). Differences were also found in the areas of activation, when taking the difference between the areas of activation in the middle of the motor cortex from the areas of activation in the contralateral hemisphere of the tapping hand. Areas were found which did not have activation according to the time-averaged reconstructed images, but had similar temporal responses as that of the activation area. Images were produced to show these areas of similarity. The same metrics used for the time-averaged images were also found for the similarity images. The distance from center of the similar areas in the ipsilateral hemisphere of the tapping hand were found to be significantly different (p < 0.0001) between normal subjects (6.20 cm +/- 2.35 cm) and cerebral palsy subjects (0.50 cm +/- 1.27 cm). The distance from center of similar areas in the controlateral hemisphere of the tapping hand were also found to be significantly different (p = 0.0183) between normal subjects (3.10 cm +/- 1.45 cm) and cerebral palsy subjects (1.00 cm +/- 2.11 cm). A significant difference (p = 0.0110) was also observed between normal subjects (13.57 cm2 +/- 5.72 cm2) and cerebral palsy subjects (3.00 cm2 +/- 10.30 cm2) in the areas of similarity, when taking the difference between the areas of similarity in the middle of the motor cortex from the areas of similarity in the controlateral hemisphere of tapping. The significant difference between normal and cerebral palsy subjects of these temporal and spatial metrics show that fNIRS can be used in the assessment of plasticity of the motor cortex in cerebral palsy patients. These temporal and image metrics can potentially be used as biomarkers to help assess treatment and improvement of CP patients in future studies. These can also be correlated with current cerebral palsy classification schemes, and further improve the sensitivity of current classification schemes.

Disciplines

Biomedical Engineering and Bioengineering | Engineering

Comments

Degree granted by The University of Texas at Arlington

Download

Share

COinS