A novel class of analog BioElectronics is introduced for the systematic implementation of ultra-low power microelectronic circuits, able to compute nonlinear biological dynamics. This class of circuits is termed ``CytoMimetic Circuits'', in an attempt to highlight their actual function, which is mimicking biological responses, as observed experimentally. Inspired by the ingenious Bernoulli Cell Formalism (BCF), which was originally formulated for the modular synthesis and analysis of linear, time-invariant, high-dynamic range, logarithmic filters, a new, modified mathematical framework has been conceived, termed Nonlinear Bernoulli Cell Formalism (NBCF), which forms the core mathematical framework, characterising the operation of CytoMimetic circuits. The proposed nonlinear, transistor-level mathematical formulation exploits the striking similarities existing between the NBCF and coupled ordinary differential equations, typically appearing in models of naturally encountered biochemical systems. The resulting continuous-time, continuous-value, low-power CytoMimetic electronic circuits succeed in simulating with good accuracy cellular and molecular dynamics and found to be in very good agreement with their biological counterparts. They usually occupy an area of a fraction of a square millimetre, while consuming between hundreds of nanowatts and few tenths of microwatts of power. The systematic nature of the NBCF led to the transformation of a wide variety of biochemical reactions into nonlinear Log-domain circuits, which span a large area of different biological model types. Moreover, a detailed analysis of the robustness and performance of the proposed circuit class is also included in this thesis. The robustness examination has been conducted via post-layout simulations of an indicative CytoMimetic circuit and also by providing fabrication-related variability simulations, obtained by means of analog Monte Carlo statistical analysis for each one of the proposed circuit topologies. Furthermore, a detailed mathematical analysis that is carefully addressing the effect of process-parameters and MOSFET geometric properties upon subthreshold translinear circuits has been conducted for the fundamental translinear blocks, CytoMimetic topologies are comprised of. Finally, an interesting sub-category of Neuromorphic circuits, the ``Log-Domain Silicon Synapses'' is presented and representative circuits are thoroughly analysed by a novel, generalised BC operator framework. This leads to the conclusion that the BC operator consists the heart of such Log-domain circuits, therefore, allows the establishment of a general class of BC-based silicon synaptic circuits, which includes most of the synaptic circuits, implemented so far in Log-domain.

This PhD aimed at characterising mechanical, structural and compositional properties of cortical bone at the micro- and nano-scale employing various techniques applied to mouse models of ageing and disease. Nanoindentation was used to probe bone at the micro-scale. Nanoindentation properties of the same control mouse bones were measured using a range of preparation, testing and analysis options. This was the first time that such systematic study was conducted. It was concluded that although the measured values were influenced by experimental and analysis choices, nanoindentation was capable of capturing relative trends in bone's mechanical properties. Nanoindentation was then coupled to a poroelastic approach to measure age-related changes in mouse bone permeability. Permeability is key to understanding fluid flow in bone, which may indicate how bone cells sense changes in the mechanical environment. These first permeability measurements in mouse bone demonstrated that the permeability caused by fluid flowing through bone's lacunar-canalicular porosity decreases with age. Porosity is expected to also affect bone's ability to resist fracture at the micro-scale. The influence of intra-cortical porosity on crack propagation was explored via extended finite element methods. A novel technique was suggested to propagate cracks through holes and applied to 2D models of the porosity of osteogenesis imperfecta mouse bone. Results showed that vascular canals affect crack propagation and might contribute to the brittleness of osteogenesis impefecta bone. Skeletal pathologies often cause alterations in bone's building blocks leading to deteriorated whole-bone toughness. Mineral properties of brittle and ductile mouse bone (models of osteogenesis imperfecta and rickets respectively) were evaluated. Results revealed that deviations in size, composition and organisation of bone mineral reduce bone's mechanical integrity both in brittle and ductile pathologic bone. The outcomes of this thesis provide a deeper understanding of bone material, which is required for future improvements in treatments for skeletal diseases.

Clinical decision making is a ubiquitous and frequent task physicians make in their daily clinical practice. Conventionally, physicians adopt a cognitive predictive modelling process (i.e. knowledge and experience learnt from past lecture, research, literature, patients, etc.) for anticipating or ascertaining clinical problems based on clinical risk factors that they deemed to be most salient. However, with the inundation of health data and the confounding characteristics of diseases, more effective clinical prediction approaches are required to address these challenges. Approximately a few century ago, the first major transformation of medical practice took place as science-based approaches emerged with compelling results. Now, in the 21st century, new advances in science will once again transform healthcare. Data science has been postulated as an important component in this healthcare reform and has received escalating interests for its potential for 'personalizing' medicine. The key advantages of having personalized medicine include, but not limited to, (1) more effective methods for disease prevention, management and treatment, (2) improved accuracy for clinical diagnosis and prognosis, (3) provide patient-oriented personal health plan, and (4) cost containment. In view of the paramount importance of personalized predictive models, this thesis proposes 2 novel learning algorithms (i.e. an immune-inspired algorithm called the Evolutionary Data-Conscious Artificial Immune Recognition System, and a neural-inspired algorithm called the Artificial Neural Cell System for classification) and 3 continuum-based paradigms (i.e. biological, time and age continuum) for enhancing clinical prediction. Cardiovascular disease has been selected as the disease under investigation as it is an epidemic and major health concern in today's world. We believe that our work has a meaningful and significant impact to the development of future healthcare system and we look forward to the wide adoption of advanced medical technologies by all care centres in the near future.

Motivation is a fundamental driving force in animal behaviour. As such, it is vital that we are aware of its influence in shaping the behaviour examined in a laboratory setting. We set out to explore how changes in motivation affected the performance of a simple visual task using both behavioural analyses and multi-unit recordings in primary visual cortex (V1) of behaving mice. To address the question we trained water-restricted mice to selectively lick for a water reward during a 'Go' stimulus in a Go/NoGo version of a two-category discrimination. This simple paradigm is widely used across the neuroscience community due its relative ease of implementation. On the level of behavioural analysis we show that depending on motivation, performance of the task is reliant on a balance of different behavioural processes. Specifically, we demonstrate the novel result that in a state of overmotivation, goal-directed instrumental contingencies are 'masked' while Pavlovian processes dominate over action selection. Secondly, we investigated the multi-unit neural correlates of a simple visual discrimination task recorded in V1. We hypothesized that the typical changes to motivational state within a behavioural session would affect sensory processing, and that the neural signature of this effect could be found in V1. Our choice of time frame for the focus of our analysis brings valuable insights to an increasingly encountered behavioural paradigm. Furthermore, we show novel evidence of temporally specific correlates of motivation and choice behaviour in mouse V1. The recent shift to experiments using head-fixed, behaving rodents by many neurophysiologists makes a thorough understanding of underlying processes paramount. Exploring the boundary between intentional choices and overt interference by motivational states cannot be underestimated. Throughout this work we aim to incorporate existing behavioural psychology literature within the context of modern neuroscience approaches; thus providing a valuable resource for both of these communities.

Synthetic Biology is an approach to the development of biological systems based on engineering principles. By using concepts from other engineering disciplines such as abstraction it is possible to break down complicated biological functions into components termed 'BioParts'. BioParts can be assembled collectively into modules and systems to carry out advanced functions, designed from the bottom up. A key part of this approach is the standardisation of BioParts and practices to aid design, testing and implementation. An automated characterisation methodology focused primarily on promoter BioParts has been developed which is potentially scalable to other BioPart families. The standardised workflow is optimised to enable BioPart characterisation under highly reproducible growth conditions, reliably producing high quality data. It has been designed around automation equipment which should ensure accurate reproduction of experiments at other sites. The automated characterisation workflow has been demonstrated to produce high quality data for both constitutive and inducible promoters. The entire Anderson promoter collection has been characterised and high details results for all library members are available for the first time. A pair of inducible promoter BioPart were characterised to obtain a deep data level regarding their activity in response to inducer over time. To allow the characterisation of more inducible BioParts in a shorter period of time, promoter engineering was also used to generate novel promoters which are induced by xylose. The development of the automated workflow should be a step towards the standardisation of characterisation protocols and production of large numbers of BioParts with associated high quality, reproducible characterisations. Standardisation will further aid the comparison of the data sets produced, potentially shining light on unknown interactions between BioParts and their environment and improving the ability of Synthetic Biologists to design novel biological systems from the ground up.

Blowflies are exquisite fliers and have long been established as models for sensorimotor research. Their flight performance crucially depends on both visual and mechanoreceptive feedback. Our understanding of how the nervous system integrates signals of these sensory modalities, however, leaves two important issues unresolved. First: As most experiments are performed on restrained animals, how are neural signals integrated when the animal actually moves? Second: Are neural mechanisms underlying optic-flow-based self-motion estimation modulated by mechanosensory modalities when controlling motor behaviour? To address these questions, I have designed a mobile platform and miniaturized electrophysiological recording equipment. Here, I describe details of the platform's mechanical and electronic components, test its performance, and demonstrate that it enables high quality recordings from an identified interneuron in the fly motion vision pathway, the H1-cell, which is believed to process optic-flow, generate during self-motion. I compare the recordings obtained with my novel platform to those gathered with much bigger setups, which are orders of magnitude heavier, and characterize the responses of the H1-cell under various stimulus conditions. My results are mostly in agreement with earlier work but include new findings on pattern size-dependent responses of the cell, which are in contrast to theoretical predictions based on previous behavioural and physiological studies. To investigate how the fly uses signals from different sensory modalities to control various optomotor behaviours and potentially vision-based collision avoidance, I have established a brain machine interface on a small mobile robot that uses the neural activity of the H1-cell to control the robot's motor system. In summary, I have created a platform for in vivo recordings of neural activity in flies on a small robot. This novel technology will enable further studies on multisensory integration to gain new insights into the design of fly sensorimotor pathways involved in course control and collision avoidance.

The lateral geniculate nucleus (LGN) has been increasingly recognized to actively regulate information transmission to primary visual cortex (V1). Although efforts have been devoted to study its morphological and functional features, the full array of response characteristics in mouse LGN as well as their dependency on subjective state have been relatively unexplored. To address the question we recorded from mouse LGN with multisite-electrode-arrays (MEAs). From a dataset with 185 single units, our results revealed several exceptional response features in mouse LGN. We also demonstrated that subtypes, such as ON-/OFF-centre and transient/sustained cells exhibited functionally distinctive features, which might indicate parallel projections. To further compare response features from the full extent of mouse LGN, we developed a three-dimension (3D) LGN volume through histological approach. This volume explicitly captures morphological features of mouse LGN and provides the preciseness to classify location of single neuron into the anterior/middle/posterior LGN. Based on this categorization, we showed that response features were not regionally restricted within mouse LGN. We further examined neural activity with subjects in high or low isoflurane states. The distinct features in LFPs between the two states indicated that adjusting isoflurane concentration could provide a reliable and controllable experimental model to explore the state-dependent neural activity in mouse visual system. Subsequently, our results demonstrated that properties, including response latency, contrast sensitivity and spatial frequency properties were modulated by isoflurane concentration. Our current work suggests that mouse LGN can dynamically regulate information transmission to the cortex using numerous mechanisms, including responding mode, modulation of neuronal responses according to subjects' states.

Backpacks are commonly used by students of all ages and there has been a growing concern in many countries in relation to the backpack loads carried by school children and its association with the rise in complaints of neck, shoulder and back pain. Of further concern is the work of Hestbaek et al. (2006) which has shown a correlation between experiencing back pain as an adolescent and experiencing low back pain as an adult. In recent years, a number of studies have investigated physiological and movement kinematic responses to load carriage, such as oxygen consumption, heart rate, gait pattern and trunk posture (Hong et al., 2000; Pascoe et al., 1997). However, most of the studies that focused on children carrying loads looked only at gait patterns and trunk and neck postures. None of the previous studies investigated the compensatory pelvic motions of school children due to increased loads. Also, it was reported that one of the major limitations of measuring pelvic kinematics whilst carrying a backpack was occlusion of retro-reflective markers, and consequently this limits the type of activity and subject to be measured using an optical motion tracking system. Despite the presence of a variety of models, there are still debates on their reliability and repeatability, and consequently there is no clearly defined standard or consensus. In this thesis, a novel methodology was developed to measure pelvic kinematics. Its repeatability and reliability was validated experimentally by comparing it to the most relevant previous method. The result of this experiment showed that the new method improved the repeatability, reliability and reproducibility of kinematics data of the pelvis and overcomes a number of theoretical and experimental limitations, such as marker occlusion. The validated method was used to develop a protocol to measure the pelvic kinematics in adolescents whilst carrying loaded backpacks of 17% and 25% of their body weight during different activities of daily living on the basis of a survey which was conducted to explore the average daily weight that children carry to school in the UK. The result of this experiment revealed that as the load increased to 25% of the body weight, the instability in postural control increased and significant changes in pelvic tilt and rotation were noted in almost all activities. It was revealed in this study that female and male subjects used different mechanism to compensate for the effect of a heavy backpack. It was evident that carriage of loaded backpack will result in alteration of the movement of the pelvis and may in future promote postural deviation and increase lower back pain.

Bone is a dynamic tissue that adapts its shape and material properties as a response to changes in mechanical demands. The process by which bone functional adaptation takes places consists of a complex cascade of events, entailing different levels of mechanical and biological regulation, which are still unknown or under debate in the scientific community. Mechanical cues are site-specific factors that play a crucial role in skeletal maintenance and adaptation. The present work is framed within this research area and aims to inspect a hypothetical relationship between organ-level load-induced mechanical cues and the adaptive response of cortical bone. More specifically, this thesis focuses on how loading parameters (such load frequency or the insertion of rest periods between load cycles) alter the mechanical environment in bone tissue (estimated using numerical tools) and, consequently, its adaptive response. Animal in vivo loading models allow for inducing bone functional adaptation in a controlled environment and are, therefore, a powerful tool in the study of the influence of external loading and loading parameters to bone mechanotransduction. The murine axial tibial loading model was used as platform to trigger cortical bone adaptation in mouse specimens and provide relevant biological and biomechanical information. The spatial distribution of in vivo functional adaptation in the murine model was assessed via morphological examination of ex vivo μCT scans from loaded and control tibiae of C57BL/6 mice, by contralateral comparison of cross-section geometrical properties and measurements. Calculation of the second moment of area indicated regions of bone formation along the length of the bone. A method for mapping cortical thickness was developed, consisting in the three-dimensional representation of the changes of shell thickness in the diaphysis of a long bone, allowing for explicitly describing how bone adaptation is distributed in space. The mechanical environment of the cortical shell was estimated using computational models, based on finite-element analysis generated from μCT data, to provide a full field description of mechanical fields. A novel in silico predictive model was coupled to the finite-element models. This mathematical formulation assumed that bone responds instantly to local mechanical cues in an on-off manner and that this response is integrated in time and averaged in space, resulting in a bone formation rate represented by surface displacements. Strain energy density (SED) was initially employed as stimulus to cortical bone formation. The obtained predictions were compared against the 3D cortical adaptation maps from in vivo adaptation, showing that SED is able to reproduce the spatial patterns of changes in bone shape, but has a limited contribution in the study of time-dependent parameters. Following experimental evidence that suggests that interstitial fluid flow in the lacunar-canalicular system is a stimulus for mechanoadaptation, a simplified 3D poroelastic finite-element model of a beam in bending was developed in order to simulate the behaviour of fluid flow in mouse cortical bone. This model allowed exploration of two important loading parameters that affect mechanoadaptation: load frequency and rest insertion. A range of intrinsic permeabilities found in literature from 1E-23 to 1E-18 m2 were tested, and fluid velocity was determined. Models with permeabilities down to 1E-21 m2 followed a dose-response relationship between fluid flow and sinusoidal frequency. Smaller orders of magnitude of permeability were relatively insensitive to frequency. It was found that there is a minimum time of rest between loading cycles that is required to maximise fluid motion. These findings suggest that, in addition to biological saturation, fluid flow plays a role in the enhancement of osteogenic response in load regimes that allow recovery periods between consecutive load cycles. Fluid velocity was then included as a mechanical stimulus in the developed cortical bone adaptation algorithm, in order to determine if, in addition to predicting time-dependent factors, fluid flow could reproduce in vivo spatial patterns. Equivalent predictions to the strain-based simulations were obtained. The presented cortical adaptation algorithm simulated spatial distribution of cortical adaptation with a good agreement between the in vivo measurements and our predictions. The work presented here provides novel methodological and theoretical approaches to understanding the spatial and temporal parameters of cortical bone adaptation. With a better understanding of the factors that promote bone formation, mechanical loading can be optimized to elicit a maximum osteogenic response.

Tissue optical and mechanical properties are related to tissue pathological changes. The ability to measure both tissue elasticity and its optical properties using the same hardware offers a significant advantage over existing techniques in, e.g. imaging of cancer. Therefore this thesis aims to develop a dual mode imaging system capable of noninvasively sensing local optical and mechanical properties at centimetre depths in samples. The proposed method is based on the detection of photons modulated by ultrasound and shear waves with a modified acoustic radiation force assisted ultrasound modulated optical tomography (ARF-UOT) system. Firstly the detection of the shear wave and ultrasound modulation with UOT was demonstrated at the surface of tissue mimicking phantoms. The ultrasound field or shear wave wavefront could be imaged by a single CCD exposure and analysis of local laser speckle contrast. Secondly, within tissue mimicking phantoms, while the shear waves cannot be imaged directly due to optical scattering, the propagation of a transient shear wave was tracked with global laser speckle contrast analysis. A differential method was developed to measure the local shear wave speed and quantify the elasticity of the tissue mimicking phantoms at ~cm depths. The method (SW-LASCA) was based on a modified ARF-UOT system. By generating continuous shear waves at different frequencies, the dispersion of shear wave speed was also investigated. The feasibility of the viscosity measurement was demonstrated by fitting the measured attenuation dispersion using the Voigt model. Finally, the dual mode system was explored by combining the SW-LASCA and ARF-UOT. The system was demonstrated in an optical reflection detection geometry and the scanning results of heterogeneous phantoms demonstrated the potential of the system to distinguish optical contrast, mechanical contrast and optical/mechanical contrast in a reflection detection geometry.