The aim of this research is to improve the understanding of the impact on the cochlear dynamics corresponding to surgical tools, processes and hearing implants such that these can be designed more appropriately in the future. The results suggest that enhanced performance of implants can be achieved by optimisation of the location with respect to the cochlea and have shown that robotic surgical tools used to enable precise, simplified processes can reduce harm and offer other benefits. With an ageing population, and where exposure to noise on daily basis is increased rather than industrial settings, at least two factors of age and noise, will contribute to a greater incidence of hearing loss in the population in the future. In the research a mathematical model of the passive cochlea was produced to increase understanding of the sensitivity and behaviour of the fluid, structure and pressure transients within the cochlea. The investigation has been complemented by an innovative experimental technique developed to evaluate the dynamics in the cochlear fluids while maintaining the integrity of the cochlear structure. This technique builds on the success of the state-of-the-art surgical robotic micro-drill. The micro-drill enables removal of bone tissue to prepare a consistent aperture onto the endosteal membrane within the cochlea. This is known as preparing a ‘Third window’. In this technique the motion of the exposed endosteal membrane is treated as the diaphragm element of a pressure transducer and is measured using a Micro- Scanning Laser Vibrometer operating through a microscope. There are two principal outcomes of the research: First, the approach has enabled disturbances in the cochlea to be contrasted for different surgical techniques, which it is expected to allude preferential methods in future surgery in otology. In particular it was shown that when using the robotic micro-drill to create a cochleostomy that the disturbance amplitude reduces to 1% of that experienced when using conventional drilling. Secondly, an empirically derived frequency map of the cochlea has been produced to understand how the location of implants affects maximum power transmission over the required frequency band. This has also shown the feasibility of exciting the cochlea at a third window in order to amplify cochlear response.

This thesis deals with automatic localization of thin surgical tools such as needles or electrodes in 3D ultrasound images. The precise and reliable localization is important for medical interventions such as needle biopsy or electrode insertion into tissue. The reader is introduced to basics of medical ultrasound (US) imaging. The state of the art localization methods are reviewed in the work. Many methods such as Hough transform (HT) or Parallel Integral Projection (PIP) are based on projections. As the existing PIP implementations are relatively slow, we suggest an acceleration by using a multiresolution approach. We propose to use model fitting approach which uses randomized sample consensus (RANSAC) and local optimization. It is a fast method suitable for real-time use and it is robust with respect to the presence of other high-intensity structures in the background. We propose two new shape and appearance models of tool in 3D US images. Tool localization can be improved by exploiting its tubularity. We propose a tool model which uses line filtering and we incorporated it into the model fitting scheme. The robustness of such localization algorithm is improved at the expense of additional time for pre-processing. The real-time localization using the shape model is demonstrated by implementation on the 3D US scanner Ultrasonix RP. All proposed methods were tested on simulated data, phantom US data (a replacement for a tissue) and real tissue US data of breast with biopsy needle. The proposed methods had comparable accuracy and the lower number of failures than the state of the art projection based methods.

Reliable chronic neural recording from focal deep brain structures is impeded by insertion injury and foreign body response, the magnitude of which is correlated with the mechanical mismatch between the electrode and tissue. Thin and flexible neural electrodes cause less glial scarring and record longer than stiff electrodes. However, the insertion of flexible microelectrodes in the brain has been a challenge. A novel insertion method is proposed, and demonstrated, for precise targeting deep brain structures using flexible micro-wire electrodes. A novel electrode guiding system is designed based on the principles governing the buckling strength of electrodes. The proposed guide significantly increases the critical buckling force of the microelectrode. The electrode insertion mechanism involves spinning of the electrode during insertion. The spinning electrode is slowly inserted in the brain through the electrode guide. The electrode guide does not penetrate into cortex. The electrode is inserted in the brain without stiffening it by coating with foreign material or by attaching a rigid support and hence the method is less invasive. Based on two new mechanisms, namely spinning and guided insertion, it is possible to insert ultra-thin micro-wire flexible electrodes in rodent brains without buckling. I have demonstrated successful insertion of 25 µm platinum micro-wire electrodes about 10 mm deep in rat brain. A novel micro-motion compensated ultra-thin flexible platinum microelectrode has been presented for chronic single unit recording. Since manual insertion of the proposed microelectrode is not possible, I have developed a microelectrode insertion device based on the proposed method. A low power low noise 16 channel programmable neural amplifier ASIC has been designed and used to record the neural spikes. The ability to record neural activity during insertion is a unique feature of the developed inserter. In vivo implantation process of the microelectrode has been demonstrated. Microelectrodes were inserted in the Botzinger complex of rat and long term respiratory related neural activity was recorded from live rats. The developed microelectrode has also been used to study brain activity during seizures. In-vivo experimental results show that the proposed method and the prototype insertion system can be used to implant flexible microelectrode in deep brain structures of rodent for brain studies.

Cette thèse traite de la détection automatique d’outils chirurgicaux de géométrie linéaire tels que des aiguilles ou des électrodes en imagerie ultrasonore 3D. Une localisation précise et fiable est nécessaire pour des interventions telles que des biopsies ou l’insertion d’électrode dans les tissus afin d’enregistrer leur activité électrique (par exemple dans le cortex cérébral). Le lecteur est introduit aux bases de l’imagerie ultrasonore (US) médicale. L’état de l’art des méthodes de localisation est rapporté. Un grand nombre de méthodes sont basées sur la projection comme la transformation de Hough ou la Projection Intégrale Parallèle (PIP). Afin d’améliorer l’implantation des méthodes PIP connues pour être assez lentes, nous décrivons une possible accélération par approche multirésolution. Nous proposons d’utiliser une méthode d’ajustement de modèle utilisant une approche RANSAC et une optimization locale. C’est une méthode rapide permettant un traitement temps réel et qui a l’avantage d’être très robuste en présence d’autres structures fortement échogènes dans le milieu environnant. Nous proposons deux nouveaux modèles d’apparence et de forme de l’outil dans les images US 3D. La localisation de l’outil peut être améliorée en exploitant son aspect tubulaire. Nous proposons un modèle d’outil utilisant un filtrage rehausseur de ligne que nous avons incorporé dans le schéma de recherche de modèle. La robustesse de cet algorithme de localisation est améliorée au prix d’un temps additionnel de pré-traitement. La localisation temps-réel utilisant le modèle de forme est démontrée par une implantation sur l’échographe Ultrasonix RP. Toutes les méthodes proposées on été testée sur des données de simulation US, des données de fantômes (qui sont des tissus synthétiques imitant les tissus biologiques) ainsi que sur des données réelles de biopsie du sein. Les méthodes proposées ont montré leur capacité à produire des résultats similaires en terme de précision mais en limitant d’avantage le nombre d’échecs de détection par rapport aux méthodes de l’état de l’art basées sur les projections. / This thesis deals with automatic localization of thin surgical tools such as needles or electrodes in 3D ultrasound images. The precise and reliable localization is important for medical interventions such as needle biopsy or electrode insertion into tissue. The reader is introduced to basics of medical ultrasound (US) imaging. The state of the art localization methods are reviewed in the work. Many methods such as Hough transform (HT) or Parallel Integral Projection (PIP) are based on projections. As the existing PIP implementations are relatively slow, we suggest an acceleration by using a multiresolution approach. We propose to use model fitting approach which uses randomized sample consensus (RANSAC) and local optimization. It is a fast method suitable for real-time use and it is robust with respect to the presence of other high-intensity structures in the background. We propose two new shape and appearance models of tool in 3D US images. Tool localization can be improved by exploiting its tubularity. We propose a tool model which uses line filtering and we incorporated it into the model fitting scheme. The robustness of such localization algorithm is improved at the expense of additional time for pre-processing. The real-time localization using the shape model is demonstrated by implementation on the 3D US scanner Ultrasonix RP. All proposed methods were tested on simulated data, phantom US data (a replacement for a tissue) and real tissue US data of breast with biopsy needle. The proposed methods had comparable accuracy and the lower number of failures than the state of the art projection based methods.

No description available.