Near-infrared spectroscopy (NIRS) is an emerging non-invasive brain-computer interface (BCI) modality that measures changes in hemoglobin concentrations in neurocortical tissue. Previous NIRS studies have not employed real-time feedback with online classification, a combination which would allow users to alter their mental strategy on the fly. This thesis reports the results of two online studies. The first study contrasted online classification of prefrontal hemodynamics using an artificial neural network (ANN) and a hidden Markov model-based (HMM) classifier. The second study measured the accuracy of an online linear discriminant classifier. In study 1, only the ANN classifier facilitated online classification rates greater than chance (p=0.0289). In study 2, a new feedback system and experimental protocol led to improved classification rates over those of the first study (p=5.1*10^(-5)). While control over instantaneously generated feedback in online NIRS-BCIs has been demonstrated, factors such as user frustration, mental fatigue, and restrictions on ambient lighting may compromise performance.

Individuals with limited motor control are often unable to interact with their environment. Recently, near-infrared spectroscopy (NIRS) systems have been investigated as potential brain-computer interfaces (BCI). Previous studies examined data offline, preventing users from understanding how their thoughts triggered the NIRS system. This thesis focused on understanding the short-term effects of feedback on user’s ability to learn how to control BCIs. Data were collected from control and experimental groups over seven sessions, as they performed fast singing imagery or mental arithmetic. Significant differences were observed between the control group’s results in non-feedback sessions and the experimental group’s results in feedback sessions. Qualitative results from 3 of the 10 participants suggested they had control of the feedback system. They performed the task with online accuracies of 61% - 88% in the final 2 sessions with feedback. These results suggest that continued investigation of NIRS feedback systems is warranted.

The synthetic methodology of tri- and pentamethine carbocyanines and their interactions with biomolecules will be discussed in two chapters. The first chapter describes the preparation of halogenated carbocyanine dyes that display multiple charges; furthermore, these particular compounds were examined for their ability to bind G-quadruplex DNA with selectivity over duplex DNA and have potential for developing novel chemotherapeutic agents. The second section discusses the synthetic methods utilized to prepare trimethine cyanine fluorophores. This chapter will show how varying the N-indolenyl substituients’ hydrophobicity from ethyl to phenylpropyl influences the binding to Human Serum Albumin (HSA); additionally, alternating the terminal heterocyclic moieties of the cyanine dye has a direct quantitative effect on the biomolecular interaction. These identical compounds were recognized to be structurally analogous to agents that commonly interact with Protein Arginine Methyl Transferase (PRMT) and these compounds display low IC50 values toward inhibition of PRMT1 with unique NIR imaging properties.

Near-infrared (NIR) fluorescence imaging has emerged as an attractive non-invasive approach for direct visualization of diseases which depends on the development of stable, highly specific and sensitive optical probes. The NIR region of the electronic spectrum offers a reduction in the background autofluorescence and an increase in the tissue penetration depth. Cyanine dyes have often been considered promising contrast optic agents owing to their photophysical properties. Herein the synthesis of various penta- and heptamethine benzothiazole cyanine dyes has been described and their in vivo imaging efficacy was determined. Varying functionalities on the benzothiazole aromatic ring and changing substituents on the benzothiazolium nitrogen atom reflected subsequent changes in the imaging pattern and have resulted in the development of promising brain targeting agents.

The widely available charge coupled device (CCD) and lately CMOS imaging devices have created many applications on a mass scale. However these devices are limited to wavelengths shorter than about 1 μm. Hybrid photon upconversion devices have been developed recently. The end goal is to achieve an alternative technology for imaging in the 1.5-μm region. The hybrid upconversion idea relies on the integration of a photodetector and an organic light emitting diode (OLED). Under a forward bias for the OLED, the detected signal in the Photodetector is sent to the OLED, resulting in an increase in emission at a shorter wavelength and therefore achieving optical up conversion. An OLED device can simply consists of a stack of anode, a hole transport layer (HTL), a light-emitting layer, an electron transport layer (ETL), a cathode layer, and it typically emits visible light. As each organic molecule is a topologically perfect structure, the growth of each organic layer does not require “lattice matching”, which has been the fundamental limit for inorganic semiconductor monolithic devices. Thus, integration of an OLED with a III–V compound semiconductor is a highly feasible and desirable approach for making low-cost, large-area, potentially high efficiency devices. This thesis addresses the physics, fabrication and characterization of hybrid near infrared optical upconverters and their imaging application. Firstly, one novel hybrid optical upconverter structure is presented, which substantially improves the upconversion efficiency by embedding a metal mirror. Efficient carrier injection from the inorganic photodetector to the OLED is achieved by the insertion of a thin Au metal embedded mirror at the inorganic-organic interface. The upconversion efficiency was improved by more than 100%. Secondly, the overall upconversion efficiency can be increased significantly, by introducing a gain mechanism into the Photodetector section of the upconverter. A promising option to implement gain is a heterojunction phototransistor (HPT). An InGaAs-InP HPT was integrated with an OLED, which converts 1.5-μm Infrared light to visible light with a built-in electrical gain (~94). The overall upconversion efficiency was improved to be 1.55 W/W. Thirdly, this upconversion approach can also be used to realize a pixelless imaging device. A pixelless hybrid upconversion device consists of a large-area single-mesa device, where the OLED output is spatially correlated with the input 1.5-µm scene. Only the parts receiving incoming photons will emit output photons. To achieve this functionality, photon-generated carriers must flow mainly along the layer-growth direction when injected from the InGaAs light absorption layer into OLED light emission layer. A prototype of pixelless imaging device based on an i-In0.53Ga0.47As/C60 heterojunction was demonstrated, which minimized lateral current spreading. This thesis presents experimental results of the first organic/inorganic hybrid optical amplifer and the first hybrid near infrared imaging device.

In this study, we fabricated an electrically wavelength-tunable resonant-cavity light emitting diode (RCLED). It was achieved by the combination of an AlGaInAs quantum well structure with an intra-cavity liquid crystal material. In the phase modulator layer, we used nematic liquid crystal (NLC) and cholesteric liquid crystal (CLC), respectively, comparing their difference in the infrared-spectrum. When inserting NLC, the anisotropic properties of liquid crystal enable continuous tuning of mode emission along the extraordinary direction and provide a 58 nm tuning range. The optical characteristics of this device are polarization dependent. On the contrary, the CLC-based device is polarization independent because any polarization of incident light experiences the same averaged refractive index. However, the phase difference of CLC is less than NLC, only a 41 nm tuning range of this sample. We also simulated and discussed experimental results of NLC-based RCLED. The optical pumping of the active region is realized by a CW laser at 1064 nm wavelength and observed at room temperature.

As variation in pork tenderness has increased, identification of tenderness has become an industry need. This study consisted of 1208 pork loins randomly selected to test the efficacy of four automated grading techniques. Visible and near-infrared spectroscopy (VVNIR) (350-1830 nm wavelengths), bioelectrical impedance (EI) (resistance, reactance, phase angle, and partial capacitance), pH, and CIE L*, a* and b* color space values were used to predict chemical moisture and lipid, pH, Warner-Bratzler shear force (WBSF), and Slice shear force (SSF) on 13 d aged pork loins. The means and standard deviations for WBSF were (22.95 and 5.16) and SSF were (165.49 and 58.15). Prediction was based on stepwise linear regression and partial least squares regression. VNIR, pH, and color, when in combination, had the highest R^2 (0.19 and 0.21) for the prediction of WBSF and SSF, respectively. Partial least squares regression (PLSR) was used to remove autocorrelation between VNIR values. By using PLSR, with an R ^2 value of 0.49, 100 percent of the "tender" chops were correctly classified, 93 percent of the "intermediate" chops were correctly classified, and 92 percent of the "tough" chops were correctly classified into its category for WBSF. However, SSF was much lower (R^2 = 0.24) with only correctly placing 62 percent of the "tender" chops and only 48 percent of the "intermediate" and "tough" chops. Electrical impedance, alone or in combination with other technologies, either did not improve predictability of linear regression equations (increase R^2) or of PLSR models (increase R^2). Equations and models that included EI values had low R^2. When adding EI to the regression equation involving all variables, R^2 increased slightly from 0.19 to 0.21 in predicting WBSF, and from 0.21 to 0.25 for SSF. When pH or CIE L* color space values were included in linear regression or PLSR models to predict WBSF and SSF, R^2 values increased from 0.14 to 0.19 for WBSF, and 0.14 to 0.21 for SSF. pH played a large role in predicting WBSF and SSF, along with CIE L*. Thus, for an on-line situation, use of VNIR, pH, and color could be used to predict tenderness. Utilization of VNIR alone could be effective in predicting pork tenderness (WBSF). Using EI alone, or in combination with VNIR, would not provide acceptable prediction of WBSF or SSF. Use of VNIR with pH and color would improve the ability to predict tender and intermediate pork WBSF and SSF, but the additional improvement in accuracy may not be warranted based on the cost and additional time needed when using more than one technology.

Thesis (M.S.)--University of Hawaii at Manoa, 2005. / Includes bibliographical references (leaves 48-50). Also available via World Wide Web.

Thesis (Ph.D.) -- University of Texas at Arlington, 2007.

Not available / text