For the last 20 years, several applications of electronic nose (e-nose) have been reported in the area of microbiology, environmental and agricultural monitoring or medical diagnosis. E-noses were used to detect contaminants or for quality control. However, little has been reported about complex methodological problems which are strongly linked to the e-nose performance. This thesis summarises various e-nose systems and alternatives for gas and headspace analysis, highlights the essential problems associated with e-nose analysis and explains why these devices have a potential for the detection of trace gas molecules but also why a stable and reliable analysis is not possible yet. Methodological weaknesses such as changes in mass flow rates, filter application or sampling methods are addressed. Understanding these enables analysis of serum and urine samples from cattle or badgers either naturally or experimentally infected with the zoonotic diseases caused by Mycoplasma bovis, Mannheimia haemolytica A1, Mycobacterium bovis, Mycobacterium avium ssp. paraTuberculosis and Brucella sp. The circumstances under which meaningful results can be obtained using the ST214 e-nose (Scensive Tech. Ltd., UK) are assessed which show the current limitations for discriminating between samples. Alternative methods for analysing e-nose data are mentioned and reasons are given why under the stated circumstances no straightforward multivariate statistics is possible. However, despite various difficulties, meaningful results at a group level were obtained and could be correlated with other results obtained using alternative analytical methods. This indicates the positive proof-of-principle character of this project.

Prostate cancer is a biologically heterogenous disease with considerable variation in clinical aggressiveness. Gleason grade, the universally accepted method for classification of prostate cancer, is subjective and gives limited predictive information regarding prostate cancer progression. There is a clinical need for an objective, reliable tool to help pathologists improve current prostate tissue analysis methods and better assess the malignant potential of prostate tumours. Fourier Transform Infrared (FTIR) microspectroscopy is a powerful bioanalytical technique that uses infrared light to interrogate biological tissue. The studies detailed in this thesis examine the ability of FTIR combined with multivariate analysis to discriminate between benign, premalignant and malignant prostate pathology in snap frozen, paraffinated and deparaffinated tissue. Prostate tissue was collected during and after urological procedures performed between 2005 and 2008. The tissue was analysed utilising a bench top FTIR system in point and image mapping modes. The histology under interrogation was identified by a uro- pathologist. Multivariate analysis was applied to the spectral dataset obtained. FTIR performance was evaluated. FTIR was able to reproducibly discriminate between benign and malignant prostate tissue in a pilot study. Cross validated diagnostic algorithms, constructed from the spectral dataset in this experiment, achieved sensitivities and specificities of 95% and 89% respectively. FTIR analysis of transverse paraffinated and deparaffinated radical prostatectomy sections achieved good differentiation of the benign, premalignant and malignant pathology groups. However the performance of diagnostic algorithms constructed from this dataset under cross validation was poor. The work in this thesis illustrates the potential of FTIR to provide an objective method to assist the pathologist in the assessment of prostate samples. The limitations of the technique and directions for future work are presented.

A coherence function is a measure of the correlation of two signals and may be used as a measure for functional relationship between brain areas. In studying functional relationships, referenced EEG (REEG) coherence analysis yields important new aspects of brain activities, which complement the data obtained by power spectral analysis. However, REEG-based coherence tends to show a false high value due to volume conduction from un correlated sources (VCUS). Existing signal processing methods address this issue using a Fourier coherence function of scalp Laplacian. Although this method has been proved useful to reveal correlation between EEG signals with minimum VCUS effects, it only provides frequency-domain analysis. Since EEG signals are highly non-stationary, it is more appropriate to use time-frequency methods for coherence analysis of scalp Laplacian. Thus this research applies the wavelet transform on coherence analysis of scalp Laplacian. To verify our technique, already recorded EEG data of event related potentials were obtained from a study of two large groups of alcoholic and abstinent alcoholic subjects, performing visual picture-recognition tasks. The proposed coherence method successfully detected time-frequency correlation between EEG signals with minimum VCUS effects. It showed significant spatial specificity and revealed detailed coherence patterns. Some new important results regarding time-frequency characteristics of VCUS effects on wavelet and short-time Fourier transform (STFT) coherence analysis of REEG signals were deduced. The proposed coherence method was also compared to a conventional wavelet coherence method of REEG signals in the study of coherence difference between coherences of alcoholic and abstinent alcoholic EEG signals. Results of this study provided substantial evidence that VCUS effects are not additive and therefore can not be ignored in comparison of different brain states between groups of subjects.

No description available.

In neurosciences, there is a great need to improve methods for direct imaging of functional activity in the nervous system of the brain. Such activities occur on a time scale of the order of 10 milliseconds. No portable system exists to measure these activities non-invasively with sufficient temporal resolution and with sensitivity to the deepest regions of the brain. Electrical Impedance Tomography (EIT), as a non-invasive means for fast imaging of impedance changes during neural activity, has the promise to be highly portable, to give information on the deepest regions of the brain, and is already applied routinely in other medical applications such as breast imaging. For this technique to 'follow' the human brain's processing cycle for individual stimuli, a temporal resolution of much less than 1 second is required. The sensitivity and spatial resolution of EIT can be enhanced significantly by increasing the measurementSNR. It has been suggested that an EIT system with measurement s~nsitivity of 80 dB is required for the measurement of impedance changes in the brain occurring directly due to neural function. For this application, the University of Manchester is developing a 32channel EIT system, fElTER, with temporal resolution of 100 frames per second and SNR of 80 dB, operating over the current excitation frequency range from 10 kHz to 100 kHz. In this thesis, a modular single-channel EIT system is designed to investigate ways of achieving the high performance requirements of the fEITER system. It uses digital waveform synthesis, a 16-bit DAC and subsequent reconstmction filter, to drive the improved Howland current source. The voltage is measured using an instmmentation amplifier followed by an anti-aliasing filter and a 16-bit ADC, prior to digital PSD where the amplitude and phase information of the measured voltage is obtained. In this system the digital parts are implemented in an FPGA. The design considerations to achieve the required performance related to each module are discussed prior to design, implementation and testing related to each module in a separate chapter. The resulting single-channel EIT system operates in the frequency range of 10 kHz to 100 kHz in steps of 10 kHz. The SNR perfoffilance of the developed system is 77 dB and it has the potential of achieving 100 frames per second image acquisition for the 32-channel fEITER system.

Reliable and accurate information regarding the heat distribution inside biologica.I tissue due to microwave thennal ablation is essential for the design and optimization of safe therapeutic applicators and for the development of experiments, which aim to investigate their effects. To date, very few finite element models (FEM) have been developed to describe and illustrate electric field penetration, specific absorption rate (SAR) and temperature distribution in biological tissue due to microwave ablation probes. A coupled field finite elemerit analysis model that accounts for the temperature dependent conductivity and blood perfusion changes has not, as yet, been developed for ablation type devices. Although it is widely acknowledged that accounting for temperature dependent phenomena may affect the outcome ofthese models, the effect has not yet been assessed for microwave tumour ablation (MTA). This thesis aims to develop an improved non-linear coupled field electromagnetic and thermal FE model of microwave ablation, incorporating temperature dependent electrical and thermal properties of, ex-vivo bovine liver. Measurements of temperature dependent ex-vivo bovine liver electrical permittivity are made and their effect. presented. Temperature dependent thennal properties of the tissue are limited. to temperatures below phase change. Therefore, a theoretical tissue model for going changes in phase is also presented. The FE model of microwave ablation is based on a 2.45 GHz minimally invasive dielectric loaded monopole antenna

This thesis Presents an investigation into the use of polymer gel dosimeters for analysis in targeted radionuclide therapy (TRT). The project aims to establish viability, technical issues, and potential uses of polymer gel dosimetry when incorporating radioisotopes. The thesis begins by outlining the current status of TRT and the need for accurate dosimetry. The inaccuracies of the dosimetry methods being employed are highlighted, as is the requirement of a "gold standard" method for dosimetry quality assurance. Polymer gel dosimeters are porposed as a possible means of achieving this. As a research topic in its own right an introduction into polymer gel dosimeters is given with a detailed description of their development and current uses in external beam radiotherapy. In the thesis polymer gel dosimeters are used in conjunction with MR and a detailed description of the MR techniques used is also given. Monte Carlo simulations are currently considered the best method to determine absorbed dose from a known activity distribution. A Monte Carlo user code written for EGSnrc has been developed to verify and validate polymer gel dosimetry. Details of the Monte Carlo code are given and comparisons made with published data. The accuracy of poymer gel dosimetry is susceptible to many factors and use with internal isotopes may introduce further technical issues. These are investigated in a number of small experiments and comparison made with external beam irradiation. A method for using polymer gel dosimeters with internal isotopes is developed and verified by comparing dose measurements and distributions with that calculated using the Monte Carlo code. Comparisons with various SPECT based dosimetry techniques have been made for geometrical phantoms of both uniform and non-uniform distributions. Methods for comparing and analysing 3D dose maps have also been investigated and the results from these experiments discussed. The thesis concludes with a case study, whereby polymer gel dosimeters are used for analysis of a specific patient situation. A patient synopsis is given with details of an anatomical phantom based on that patient. Polymer gel and SPECT based dosimetry is performed and the clinical significance of the findings discussed.

Electroencephalography (EEG) and magnetoencephalography (MEG), which are two of a number of neuroimaging techniques, are scalp recordings of the electrical activity of the brain. EEG and MEG (E/MEG) have excellent temporal resolution, they are easy to acquire, and have a wide range of applications in science, medicine and engineering. These valuable signals, however, suffer from poor spatial resolution and in many cases from very low signal to noise ratios. In this study, new computational methods for analyzing and improving the quality of E/MEG signals are presented. We mainly focus on single trial event-related potential (ERP) estimation and E/MEG dipole source localization. Several methods basically based on particle filtering (PF) are proposed. First, a method using PF for single trial estimation of ERP signals is considered. In this method, the wavelet coefficients of each ERP are assumed to be a Markovian process and do not change extensively across trials. The wavelet coefficients are then estimated recursively using PF. The results both for simulations and real data are compared with those of the well known Kalman Filtering (KF) approach. In the next method we move from single trial estimation to source localization of E/MEG signals. The beamforming (BF) approach for dipole source localization is generalized based on prior information about the noise. BF is in fact a spatial filter that minimizes the power of all the signals at the output of the filter except those that come from the locations of interest. In the proposed method, using two more constraints than in the classical BF formulation, the output noise powers are minimized and the interference activities are stopped.

A spectral-domain OCT system operating at 1300 nm wavelength region, capable of acquiring 47,000 A-lines/s, was designed and developed. Its axial and transverse resolutions were ∼ 6 µm and ∼15 µm respectively. OCT images of human skin were obtained in vivo using three OCT systems, in order to find the optimal wavelength region for dermal imaging. 800 nm OCT system provided better image contrast over other two wavelength regions. Meanwhile, 1300 nm wavelength region was needed to obtain information from deeper dermal layers. To determine the effect of melanin pigmentation on OCT, images were taken from subjects with different ethnic origins. Interestingly, melanin pigmentation was found to have little effect on penetration depth in OCT. In vitro tumour samples, comprising samples with different degrees of dysplasia, were imaged at 800 nm, 1060 nm and 1300 nm wavelength regions to find the capability of OCT to diagnose microstructural changes occurring during tumour progression. 800 nm OCT system was capable to detect the malignant changes with higher contrast than other wavelength regions. However, higher wavelength regions were required to penetrate deeper in densely scattering tumour samples at advanced stages. OCT system operating at 1060 nm was combined with a photoacoustic imaging (PAT) system to obtain complementary information from biological tissues. This multimodal OCT/PAT system demonstrated its potential to deliver microstructural information based on optical scattering and vascular information based on optical absorption in living mice and human skin. The results indicate OCT as a promising imaging modality that can have profound applications in several areas of clinical diagnostic imaging.

The aim of this study was to determine if photoplethysmography (PPG) could be used to analyse the foot microvascular changes caused by diabetic autonomic neuropathy. The digital PPG signals were collected from 37 healthy volunteers (Group I), 35 diabetic patients (Group II), and 38 diabetic patients with sensory neuropathy (Group III) and analysed using MAT LAB. Prominent spectral peaks with sidebands were obtained at both the high frequency (HF) and the low frequency (LF) end of the Fourier spectrum of these PPG signals. Previous studies of microcirculation have shown that both are sympathetically and parasympathetically mediated and hence are a good measure of the autonomic activity. In the HF analysis, the heart rate (HR) response from 13 participants in Group III was severely reduced and significantly different from the responses obtained from the other two groups. However the responses from remaining 25 participants had similar characteristics to those of Group II. Hence the HF analyses failed to both statistically and objectively differentiate between the diabetics with and without neuropathy. The spectral density for the frequency bandwidth of 3-20 cpm was significantly reduced in the neuropathic group, compared to the other two groups. A Statistically significant difference was observed in the spectral densities calculated from Group II and III, though no difference could be established between Groups I and III. The LF analysis of this bandwidth differentiated between Groups II and III with a sensitivity of 84% and specificity of 61%. Activities at the LF end of the spectrum mostly represent the sympathetic control as opposed to the HR variability that is mostly a measure of the parasympathetic control. These results suggest that sympathetic dysfunction possibly precedes parasympathetic dysfunction and that PPG can assess the changes in the skin microcirculation due to sympathetic damage with moderate success.