Background. A visual prosthesis is a conceptual device designed to activate residual functional neurons in the visual pathway of blind individuals to produce artificial vision. Such device, when applied to stimulate the vitreous surface of the retina, has proven feasible in producing patterned light perception in blind individuals suffering from dystrophic diseases of the retina, such as aged-related macular degeneration (AMD). However the practicality of such approach has been challenged by the difficulty of surgical access and the risks of damaging the neuroretina. Positioning a visual implant over the scleral surface of the eye could present a safer alternative but this stimulation modality has not been tested in diseased retinas. Additionally, recent research has shown that the adult neocortex retains substantial plasticity following a disruption to its visual input and the potential deterioration in visual capabilities as a result of such experience modification may undermine the overall bionic rescue strategy. Methods. Two animal models mimicking the principal pathologies found in AMD, namely photoreceptor degeneration and reduced retinal ganglion cell mass, were used to evaluate the efficacy of trans-scleral stimulation of the retina by recording electrical evoked potentials in the visual cortex. The visual performance following the loss of pattern vision induced by bilateral eyelid suturing in adult mice was examined by analysing visual evoked potentials. Findings. Spatially differentiated cortical activations were obtained notwithstanding the underlying retinopathy in the experiment animals. The charge density thresholds were found to be similar to controls and below the bioelectric safety limit. After prolonged visual deprivation (weeks) in the mouse, the visual cortical responses evoked by either electrical or photic stimuli were both significantly reduced. An assessment of different visual capabilities using patterned stimuli demonstrated that whilst visual acuity and motion sensitivity were preserved, significant depression in luminance and contrast sensitivities was detected. Conclusion. Trans-scleral stimulation of the retina is a feasible approach for the development of a visual prosthesis. Following visual loss the adult brain exhibits significant experience-dependent modifications. These new insights may force a revision on the current bionic rescue strategy.

Background. A visual prosthesis is a conceptual device designed to activate residual functional neurons in the visual pathway of blind individuals to produce artificial vision. Such device, when applied to stimulate the vitreous surface of the retina, has proven feasible in producing patterned light perception in blind individuals suffering from dystrophic diseases of the retina, such as aged-related macular degeneration (AMD). However the practicality of such approach has been challenged by the difficulty of surgical access and the risks of damaging the neuroretina. Positioning a visual implant over the scleral surface of the eye could present a safer alternative but this stimulation modality has not been tested in diseased retinas. Additionally, recent research has shown that the adult neocortex retains substantial plasticity following a disruption to its visual input and the potential deterioration in visual capabilities as a result of such experience modification may undermine the overall bionic rescue strategy. Methods. Two animal models mimicking the principal pathologies found in AMD, namely photoreceptor degeneration and reduced retinal ganglion cell mass, were used to evaluate the efficacy of trans-scleral stimulation of the retina by recording electrical evoked potentials in the visual cortex. The visual performance following the loss of pattern vision induced by bilateral eyelid suturing in adult mice was examined by analysing visual evoked potentials. Findings. Spatially differentiated cortical activations were obtained notwithstanding the underlying retinopathy in the experiment animals. The charge density thresholds were found to be similar to controls and below the bioelectric safety limit. After prolonged visual deprivation (weeks) in the mouse, the visual cortical responses evoked by either electrical or photic stimuli were both significantly reduced. An assessment of different visual capabilities using patterned stimuli demonstrated that whilst visual acuity and motion sensitivity were preserved, significant depression in luminance and contrast sensitivities was detected. Conclusion. Trans-scleral stimulation of the retina is a feasible approach for the development of a visual prosthesis. Following visual loss the adult brain exhibits significant experience-dependent modifications. These new insights may force a revision on the current bionic rescue strategy.

The electrical stimulation of appropriate components of the human visual system can result in the perception of blobs of light (or phosphenes) in totally blind patients. By stimulating an array of closely aligned electrodes it is possible for a patient to perceive very low-resolution images from spatially aligned phosphenes. Using this approach, a number of international research groups are working toward developing multiple electrode systems (called Artificial Human Vision (AHV) systems or visual prostheses) to provide a phosphene-based substitute for normal human vision. Despite the great promise, there are currently a number of constraints with current AHV systems. These include limitations in the number of electrodes which can be implanted and the perceived spatial layout and display frequency of phosphenes. Therefore the development of computer vision techniques that can maximise the visualisation value of the limited number of phosphenes would be useful in compensating for these constraints. The lack of an objective method for comparing different AHV system displays, in addition to comparing AHV systems and other blind mobility aids (such as the long cane), has been a significant problem for AHV researchers. Finally, AHV research in Australia and many other countries relies strongly on theoretical models and animal experimentation due to the difficult of prototype human trials. Because of this constraint the experiments conducted in this thesis were limited to simulated AHV devices with normally sighted research participants and the true impact on blind people can only be regarded as approximated. In light of these constraints, this thesis has two general aims. The first aim is to investigate, evaluate and develop effective techniques for mobility assessment which will allow the objective comparison of different AHV system phosphene presentation methods. The second aim is to develop a useful display framework to guide the development of AHV information presentation, and use this framework to guide the development of an AHV simulation device. The first research contribution resulting from this work is a conceptual framework based on literature reviews of blind and low vision mobility, AHV technology, and computer vision. This framework incorporates a comprehensive number of factors which affect the effectiveness of information presentation in an AHV system. Experiments reported in this thesis have investigated a number of these factors using simulated AHV with human participants. It has been found that higher spatial resolution is associated with accurate walking (reduced veering), whereas higher display rate is associated with faster walking speeds. In this way it has been demonstrated that the conceptual framework supports and guides the development of an adaptive AHV system, with the dynamic adjustment of display properties in real-time. The second research contribution addresses mobility assessment which has been identified as an important issue in the AHV literature. This thesis presents the adaptation of a mobility assessment method from the blind and low vision literature to measure simulated AHV mobility performance using real-time computer based analysis. This method of mobility assessment (based on parameters for walking speed, obstacle contacts and veering) is demonstrated experimentally in two different indoor mobility courses. These experiments involved sixty-five participants wearing a head-mounted simulation device. The final research contribution in this thesis is the development and evaluation of an original real-time looming obstacle detector, based on coarse optical flow, and implemented on a Windows PocketPC based Personal Digital Assistant (PDA) using a CF card camera. PDA based processors are a preferred main processing platform for AHV systems due to their small size, light weight and ease of software development. However, PDA devices are currently constrained by restricted random access memory, lack of a floating point unit and slow internal bus speeds. Therefore any real-time software needs to maximise the use of integer calculations and minimise memory usage. This contribution was significant as the resulting device provided a selection of experimental results and subjective opinions.

Ce travail se situe dans le cadre d'un projet pluridisciplinaire visant à développer une prothèse visuelle. La technique la plus utilisée actuellement dans de nombreux types de neuroprothèses est basée sur l'excitation par voie électrique via des électrdes. Les inconvénients d'une telle technique (très invasive, de faible résolution spatiale et par contact) pourraient être surmontés en utilisant une stimulation par laser infra-rouge. Nous présentons dans un premier temps les caractéristiques des trois diodes lasers fibrés émettant à 1875 nm, 1535 nm et 1470 nm pour des gammes de puissances optiques de quelques centaines de mW qui ont été utilisés et intégrés à deux dispositifs de mesures permettant l'observations de variations d'échanges ioniques transmembranaires (imagerie de fluorescence des ions calciums et mesure électrophysiologique par la technique de patch clamp). Nous montrons ensuite que des réponses biologiques ont été obtenues par les trois lasers, non seulement sur des cellules ganglionnaires de la rétine et du vestibule de culture mais aussi sur des tranches de rétine. L'influence des paramètres clés comme la longueur d'onde, la durée de stimulation, les seuils d'énergie a été étudié, et a permis d'établir que les seuils d'énergie de stimulation dépendent de la valeur du coefficient d'absorption de l'eau qui varie suivant la longueur d'onde utilisée. Enfin, une étude est consacrée pour expliquer les mécanismes physiques et biologiques apparaissant au cours de l'interaction du laser avec le neurone au niveau cellulaire. Des simulations numériques quantifiant l'élévation de température associées à des tests pharmacologiques cherchant à déterminer la nature des canaux ioniques spécifiques mis en jeu suggèrent la prédominance d'un effet thermique. / This work is part of a pluridisciplinary project, aiming at developing a visual prosthesis. The most used technique for this kind of neuroprosthesis is based on the electrical stimulation of nerves by electrodes. Drawbacks of such a technique (very intrusive, low spatial resolution and physical contact) could be overcome by the use of an infra red laser based stimulation. We present first the three fibre pigtailed laser diode characteristics emitting few hundred of mW at 1875 nm, 1535 nm and 1470 nm. These lasers have been integrated on two measurement devices (a fluorescence microscope and a microscope using patch clamp recording), for the observation of ionic membrane exchanges. Our results show that action potentials have been obtained by laser stimulation from the three lasers, both on retinal or vestibular ganglion cells from mass cultures and on retinal slices. The effect of key parameters as the wavelength, the stimulation time, the energy thresholds has been studied and show that the energy thresholds clearly depend on the absorption coefficient of water which varies with the wavelength. Finally, we present the results of a preliminary study aiming at determining the biophysical interaction mechanisms at cell level. Numerical simulations giving the local increase of temperature and tests of specific blocking molecules in order to know the exact nature of the ionic channels involved suggest a predominant thermal mechanism.

Ces travaux s'inscrivent dans le cadre plus général de la réalisation d'une prothèse visuelle, destinée aux personnes atteintes de cécité tardive due à une maladie dégénérative des cellules photo réceptrices de la rétine. Par la stimulation adéquate de certaines parties des voies optiques encore fonctionnelles, il est possible de transmettre à nouveau des perceptions visuelles, appelées phosphènes, dans le champ visuel du patient. Les études actuelles portant sur le contenu informationnel de cette vision prothétique proposent simplement de réduire la résolution de l'image d'une caméra embarquée. Notre étude propose une nouvelle approche basée sur le principe de la vision fonctionnelle, développée en collaboration avec les spécialistes de la rééducation fonctionnelle de l'institut ARAMAV. Par le biais de cette approche innovante, nous avons défini des méthodes d'extraction et de représentation des informations de la scène. L'objectif est d'améliorer l'autonomie en mobilité et de rendre possible la perception des expressions faciales. Pour évaluer la pertinence de ces propositions, nous avons développé un simulateur de vision prothétique, dont les paramètres sont basés sur les résultats d'essais cliniques. Pour la mobilité, nous mettons notamment en évidence l'intérêt de fournir un accès aux informations 3D, en substitution ou en complément de l'information de luminosité, ainsi que l'intérêt d'utiliser un oculomètre pour améliorer et faciliter la prise d'informations. Des expériences spécifiques à la mobilité, ainsi que certains résultats initiaux sont présentés. Pour la reconnaissance des expressions faciales, notre étude apporte la preuve que ces expressions sont correctement perçues si nous combinons des techniques de sélection de l'information et l'utilisation de certains prétraitements. / This work falls within the broader framework of visual prostheses conception, designed for people suffering from late blindness due to degenerative diseases of retina photoreceptor cells. By stimulating certain part of the optical pathway that is still functional, it is possible to elicit visual perceptions, called phosphenes, in the subject's visual field. Recent studies on the problematic of informational content of prosthetic vision propose, in majority, a simple reduction in the resolution of grayscale images acquired from a single head worn camera. Our study proposes a new approach based on the principle of functional vision, developed in collaboration with specialists in functional rehabilitation of the ARAMAV institute. Through this innovative approach, we have defined methods for extraction and representation of scene informational content, which aim to improve autonomy in mobility and to make possible the perception of facial expressio ns. In order to evaluate these propositions, we have developed a prosthetic vision simulator, whose parameters are based on clinical trials results. For mobility, we present the usefulness of providing access to 3D information, in substitution or in addition with brightness information and the usefulness of an eye tracking device to improve and facilitate spatial knowledge acquisition. Experiments dedicated to mobility as well as some preliminary results are presented. For the recognition of facial expressions, our study provides evidence that these expressions are correctly perceived if we combined information selection and specific image processing techniques.

In recent years, probabilistic models have become fundamental techniques in machine learning. They are successfully applied in various engineering problems, such as robotics, biometrics, brain-computer interfaces or artificial vision, and will gain in importance in the near future. This work deals with the difficult, but common situation where the data is, either very noisy, or scarce compared to the complexity of the process to model. We focus on latent variable models, which can be formalized as probabilistic graphical models and learned by the expectation-maximization algorithm or its variants (e.g., variational Bayes).<br> After having carefully studied a non-exhaustive list of multivariate kernel density estimators, we established that in most applications locally adaptive estimators should be preferred. Unfortunately, these methods are usually sensitive to outliers and have often too many parameters to set. Therefore, we focus on finite mixture models, which do not suffer from these drawbacks provided some structural modifications.<br> Two questions are central in this dissertation: (i) how to make mixture models robust to noise, i.e. deal efficiently with outliers, and (ii) how to exploit side-channel information, i.e. additional information intrinsic to the data. In order to tackle the first question, we extent the training algorithms of the popular Gaussian mixture models to the Student-t mixture models. the Student-t distribution can be viewed as a heavy-tailed alternative to the Gaussian distribution, the robustness being tuned by an extra parameter, the degrees of freedom. Furthermore, we introduce a new variational Bayesian algorithm for learning Bayesian Student-t mixture models. This algorithm leads to very robust density estimators and clustering. To address the second question, we introduce manifold constrained mixture models. This new technique exploits the information that the data is living on a manifold of lower dimension than the dimension of the feature space. Taking the implicit geometrical data arrangement into account results in better generalization on unseen data.<br> Finally, we show that the latent variable framework used for learning mixture models can be extended to construct probabilistic regularization networks, such as the Relevance Vector Machines. Subsequently, we make use of these methods in the context of an optic nerve visual prosthesis to restore partial vision to blind people of whom the optic nerve is still functional. Although visual sensations can be induced electrically in the blind's visual field, the coding scheme of the visual information along the visual pathways is poorly known. Therefore, we use probabilistic models to link the stimulation parameters to the features of the visual perceptions. Both black-box and grey-box models are considered. The grey-box models take advantage of the known neurophysiological information and are more instructive to medical doctors and psychologists.<br>