High levels of random noise are a defining characteristic of neurological signals at all levels, from individual neurons up to electroencephalograms (EEG). These random signals degrade the performance of many methods of neuroengineering and medical neuroscience. Understanding this noise also is essential for applications such as real-time brain-computer interfaces (BCIs), which must make accurate control decisions from very short data epochs. The major type of neurological noise is of the so-called 1/f-type, whose origins and statistical nature has remained unexplained for decades. This research provides the first simple explanation of 1/f-type neurological noise based on biophysical fundamentals. In addition, noise models derived from this theory provide validated algorithm performance improvements over alternatives. Specifically, this research defines a new class of formal latent-variable stochastic processes called hidden quantum models (HQMs) which clarify the theoretical foundations of ion channel signal processing. HQMs are based on quantum state processes which formalize time-dependent observation. They allow the quantum-based calculation of channel conductance autocovariance functions, essential for frequency-domain signal processing. HQMs based on a particular type of observation protocol called independent activated measurements are shown to be distributionally equivalent to hidden Markov models yet without an underlying physical Markov process. Since the formal Markov processes are non-physical, the theory of activated measurement allows merging energy-based Eyring rate theories of ion channel behavior with the more common phenomenological Markov kinetic schemes to form energy-modulated quantum channels. These unique biophysical concepts developed to understand the mechanisms of ion channel kinetics have the potential of revolutionizing our understanding of neurological computation. To apply this theory, the simplest quantum channel model consistent with neuronal membrane voltage-clamp experiments is used to derive the activation eigenenergies for the Hodgkin-Huxley K+ and Na+ ion channels. It is shown that maximizing entropy under constrained activation energy yields noise spectral densities approximating S(f) = 1/f, thus offering a biophysical explanation for this ubiquitous noise component. These new channel-based noise processes are called generalized van der Ziel-McWhorter (GVZM) power spectral densities (PSDs). This is the only known EEG noise model that has a small, fixed number of parameters, matches recorded EEG PSD's with high accuracy from 0 Hz to over 30 Hz without infinities, and has approximately 1/f behavior in the mid-frequencies. In addition to the theoretical derivation of the noise statistics from ion channel stochastic processes, the GVZM model is validated in two ways. First, a class of mixed autoregressive models is presented which simulate brain background noise and whose periodograms are proven to be asymptotic to the GVZM PSD. Second, it is shown that pairwise comparisons of GVZM-based algorithms, using real EEG data from a publicly-available data set, exhibit statistically significant accuracy improvement over two well-known and widely-used steady-state visual evoked potential (SSVEP) estimators.

<p>Loss of bone mass has long been recognized as a major factor which makes bones brittle and susceptible to fracture. Currently bone mass is measured using a dual energy photon transmission technique, and a fracture risk is derived from comparison with reference normal values. Although the risk of fracture increases as bone mass decreases, variations in trabecular bone architecture can also affect strength. Consequently, trabecular bone architecture is often cited as a factor which might contribute significantly to fracture risk. Currently, estimates of trabecular bone structure are derived from biopsy studies. Such studies are invasive, destructive, cannot be used routinely in patients ar volunteers, and certainly cannot be repeated at the same site to obtain longitudinal measurements. If routine clinical assessments of architecture are to be made, it is necessary to determine which imaging modality best reveals structure in a non-invasive manner. It is also necessary to determine how the competence of the structure can best be expressed quantitatively. This work has examined ways of assessing trabecular bone structure at the distal radius in-vivo to better understand the contribution of architecture to fracture risk. To this end, it proceeded on four major fronts. First, images of sufficient resolution were acquired using a commercial pQCT scanner and a clinical MR imager. Second, the image processing software necessary to segment the imaged trabecular structure was developed. Third, two indices were proposed to quantify the connectivity of the segmented structure. One index was derived from the application of trabecular strut analysis to a skeletonized representation of the bone network. The other quantified the marrow space by deriving a mean hole area and maximum hole area of the bone structure as it appears in two dimensions. The clinical value of these indices was tested by conducting pilot studies which examined the ability of the indices to discriminate a small group of Colles fracture patients from the normal population and to reflect normal age related changes in structure. The proposed structural parameters better discriminated Colles' fracture patients than did measures of bone mineral density. The fourth and last stage of this work examined the proportion of the variance in compressive strength of a group of radius bones that can be accounted for by bone mineral density and bone architecture. In seeking the features that were the most reliable indicators of bone strength, a combination of the mean hole area and maximum hole area had the highest correlation with peak load at fracture. This held true whether these two variables were derived from pQCT or MR images. Therefore, these structural indices may represent a potentially exciting and promising means of discriminating fracture outcomes and monitoring changes in trabecular bone structure.</p> / Thesis / Doctor of Philosophy (PhD)

Thesis (Ph.D.)--University of Illinois at Urbana-Champaign, 2008. / Source: Dissertation Abstracts International, Volume: 69-11, Section: B, page: 6638. Adviser: Yi Lu. Includes bibliographical references. Available on microfilm from Pro Quest Information and Learning.

<p>The purpose of this study is to assess the feasibility of a non-invasive, reliable, and cheap method to evaluate iron-overload in beta-thalassemia major patients. The approach taken was through the possibility of in-vivo measurement of iron in the skin using a technique called x-ray fluorescence. It was hoped that the quantification of iron levels in the skin will correlate with those levels in major parenchymal organs, such as the liver and the heart – where most iron deposition occurs in thalassemic patients. Water phantoms were used to produce a calibration line with an R² value of 0.998. Skin, liver and heart tissues from 36 control mice were measured and their iron levels quantified. Iron concentration range in the skin was found to be -2 – 38 ppm with an average of 9.8 ± 1.6 ppm. Significant correlation was found between the iron levels in skin vs. heart (R_s² of 0.382); however, it was not significant in skin vs. liver (R_s² of 0.080). Skin biopsies from various sites of 6 cadavers were investigated in a synchrotron light source facility. Maps of iron, zinc and calcium distribution as a function of skin depth were then constructed. It was found that all three elements were significantly present in the epidermal layer compared to the dermal one. Calcium and zinc were present in the entire epidermis, whereas iron was mainly concentrated at the deepest region of the epidermis. It was also concluded that skin samples from the back, arm and thigh gave the clearest elemental distribution.</p> / Master of Science (MSc)

Thesis (Ph.D.)--University of Illinois at Urbana-Champaign, 2009. / Source: Dissertation Abstracts International, Volume: 70-06, Section: B, page: 3528. Adviser: Deborah E. Leckband. Includes bibliographical references (leaves 143-160) Available on microfilm from Pro Quest Information and Learning.

<p>Generation of narrow segments is a matter of concern in step-and-shoot intensity modulated radiotherapy for several reasons. The measurement, calculation and delivery of dose from narrow segments may be complicated due to: the dosimetric properties of the detector; the effect of beam penumbra and heterogeneities within the patient; and the requirement for high geometric delivery precision respectively. The main purpose of this thesis was to investigate the parameters affecting the generation of narrow beam segments in IMRT optimization. Parameters such as effective source size, Gaussian height and width, density of the target volume, and gap between the tumor and normal tissue were varied to determine their influence on the number of narrow leaf pair separations. The gradient and penumbra were also examined. Two simple geometric models (thick model and thin model) with different dimensions were used. In the thick model, two 6-MV photon beams were incident on the target at right angles. A rectangular target was centered in a phantom with dimensions 20.25 cm×5.25 cm×20.25 cm. In the thin model, one 6-MV photon beam was normally incident on a 20.25 cm×1.25 cm×20.25 cm slab phantom. The relationship between the penumbra and number of narrow separated leaf pairs were examined for the thick model. The results did not show a consistent pattern. For the thin model, creating a gap between the target and the OAR decreased the total number of narrowly separated leaf pairs along the interface but increased the average dose delivered to the OAR. By varying the OAR max dose or the gap between the target and OAR, a peak was created in the dose profiles to compensate the penumbra. As gradient increased the peak height increased to compensate the dose fall-off. The width of the peak at half maximum changed with gradient but not in a predictable fashion.</p> / Master of Science (MSc)

<p>Photodynamic therapy (PDT) is a rapidly developing clinical treatment modality involving a light-activatable photosensitizer, tissue oxygen and light of an appropriate wavelength to generate cytotoxic reactive molecular species - primarily singlet oxygen (¹O₂). Singlet oxygen readily reacts with surrounding biomolecules leading to different biological effects and subsequent therapeutic outcomes. Over the last decades, many standard PDT treatments have been approved worldwide to treat different medical conditions ranging from a variety of cancer conditions to age-related macular degeneration (AMD). Meanwhile, many active clinical trials and pre-clinical studies are underway for other clinical indications. The therapeutic outcomes of PDT are difficult to predict reliably even with many years of research. The fundamental cause for this is the inherent complexity of PDT mechanisms. As PDT involves three main components, the outcomes of PDT are determined by the combination of all components. Each component varies temporally and spatially during PDT, and the variations are mutually dependent on each other. Moreover, components such as the photosensitizer can have great variations in their initial distribution among patients even before PDT treatment. Given this, no well accepted standard PDT dose metric method has been recognized in clinics, although different approaches including explicit, implicit and direct dosimetry have been studied. To tackle the inherently complicated PDT mechanism in order to provide insights into PDT and PDT dosimetry, a theoretical one-dimensional model for aminolevulinic acid (ALA) induced protoporphyrin IX (PpIX)-PDT of human skin was developed and is presented in this thesis. The model incorporates major photophysical and photochemical reactions in PDT, and calculated temporal and spatial distributions of PDT components as well as the detectable emission signals including both sensitizer fluorescence and singlet oxygen luminescence (SOL) using typical clinical conditions. Since singlet oxygen is considered to cause PDT outcomes, the correlations of different PDT dose metrics to average reacted (¹O₂) "dose" and "dose" at different depths were examined and compared for a wide range of varied treatment conditions. The dose metrics included absolute fluorescence bleaching metric (AFBM), fractional fluorescence bleaching metric (FFBM) and cumulative singlet oxygen luminescence (CSOL), and the varied treatment conditions took into account different treatment irradiances and wavelengths, varied initial sensitizer concentration and distribution, and a wide range of optical properties of tissue. These investigations and comparisons provide information about the complicated dynamic process of PDT such as the induction of tissue hypoxia, photosensitizer photobleaching and possible PDT-induced vascular responses. It was also found that the CSOL is the most robust and could serve as a gold standard for the testing of other techniques. In addition to these theoretical studies, recent progress on the assessment of a novel, more efficient superconducting nanowire single photon detector (SNSPD) for singlet oxygen luminescence detection will be introduced and the current photomultiplier tubes (PMT) system will be briefly described as well. The author participated in the experimental assessments of the SNSPD and analyzed the results shown in this thesis.</p> / Doctor of Philosophy (PhD)

<p>The image-guided radiation therapy (IGRT) protocol used at Juravinski Cancer Center for post-prostatectomy patients involves acquiring a kV cone beam computed tomography (CBCT) image at each fraction and shifting the treatment couch to align surgical clips. This IGRT strategy is promising but its dosimetric impact is unknown, it requires significant resources, and delivers non-negligible doses to normal tissues. The objective of this work is to evaluate this IGRT protocol and investigate possible alternatives.</p> <p>IGRT delivered dose is reconstructed by deforming the planning CT to the CBCT images acquired at each fraction, computing dose on the deformed images, and inversely transforming the dose back to the original geometry. The treatments of six patients were evaluated under four scenarios: no guidance (Non-IGRT), daily guidance as performed clinically (IGRT), guidance on alternating days (Alt-IGRT), and daily automated guidance (Auto-IGRT). For one patient, the impact of reducing the planning target volume (PTV) margin to five (IGRT-5) and eight (IGRT-8) mm isotropic was also evaluated.</p> <p>With the standard clinical PTV margin of ten/seven mm, the evaluated alternatives produced similar results. The minimum dose to the CTV was decreased by 1.6±1.0, 1.2±0.7, and 0.8±0.8 Gy for Non-IGRT, Alt-IGRT, and IGRT, respectively. IGRT with manual shifting did not appear to significantly improve the delivered treatment dose compared to Auto-IGRT (difference in CTV minimum dose was 1.2±2.1Gy). Doses to the organs at risk varied but in general, an increased volume of the bladder and rectum received low doses while smaller portions received high doses. The IGRT-5 and -8 analyses showed the same CTV dose can be delivered with significant reduction in normal tissue exposure. Overall, the desired doses are delivered during IGRT although much of this may be attributed to the large PTV margins currently employed clinically.</p> / Master of Science (MSc)

<p>The scintillation crystal is a critical component in positron emission tomography (PET) systems. It impacts a number of PET system performance parameters, including spatial, energy, and time resolution. Our goal is to develop a new simulation tool to achieve improved accuracy by addressing several limitations in the existing packages, including more advanced surface treatments, temporal dependency of photon arrival, and rigorous experimental validations. The comparison of preliminary Monte Carlo simulation results and analytical calculations for specular reflection suggest that the simulation model is working well. The time-resolved light output was studied for various crystal surface treatment configurations. The measured energy resolutions are in the range of approximately 10% to 15%, which are in good agreement with published literatures. Based on the simulation and experimental results, the polished surface treatment, used together with an external specular reflector, is able to provide the best energy resolution and timing resolution for a LYSO (3x3x20 mm³) and SiPM assembly we tested. The AsCut surface with external diffusion reflector is not desired due to its inferior energy and timing resolutions. The direction and recommendation of improvements of simulation regarding surface models and wavelength dependency, as well as potential optimization of experiment such as timing pickoff methods, are discussed.</p> / Master of Science (MSc)

Distant metastases are a limiting factor in cancer patient survival as they are least accessible to conventional therapies. Effective therapy should treat primary tumours and metastatic disease. Use of image-guided radiation therapy (IGRx) enables high doses of radiation to be delivered for better tumour control while minimizing toxicity to healthy tissues. Systemic effects on distant non-irradiated tissues have been observed following IGRx. This phenomenon, termed the abscopal effect, is hypothesized to be mediated by the immune system. The inflammatory milieu generated following IGRx may activate immune cells to mount specific anti-tumour responses. The work described in this thesis aims to develop a model to study the abscopal effect, and evaluate the potential of combining IGRx and immunotherapy to enhance such distant tumour killing. Results from these studies may have clinical implications, where a combined IGRx and immunotherapy approach may prove useful in eliciting regression of local tumours and distant metastases.