Medical Bionics - Theses
Permanent URI for this collection
Search Results
1 - 5 of 5
-
ItemAssessing Speech Detection and Discrimination in Sleeping Infants Using fNIRS( 2023-10)Speech discrimination is critical in the development of speech and language in infants. The existing clinical tools are inadequate to measure speech discrimination in young infants. This thesis investigates the use of functional near-infrared spectroscopy (fNIRS), a novel, objective and infant-friendly neuroimaging technique, to measure speech detection and discrimination abilities in sleeping infants. The first study showed that the heart rate data extracted from fNIRS recordings can measure speech detection and discrimination ability. This simultaneously recorded information, when combined with the fNIRS hemodynamic response, has the potential to increase the effectiveness of measuring speech discrimination using fNIRS. Subsequent studies explore the morphology of the fNIRS response patterns to the detection and discrimination of speech sounds, using a new stimulus presentation protocol. We discovered an fNIRS response pattern not previously reported, consisting of two independent responses which changed differently over time. The differences in response morphology for various speech contrasts are attributed to varying brain arousal responses. These discoveries significantly contributed to the development of an accurate fNIRS response model for inferential analysis.
-
ItemDeep brain stimulation evoked resonant neural activity( 2023-02)Deep brain stimulation (DBS) involves surgically implanting electrodes within targeted subcortical structures and applying chronic electrical stimulation to produce therapeutic effects. DBS has utility for a range of movement and psychiatric disorders, with DBS of the subthalamic nucleus (STN) for the treatment of Parkinson’s disease a common application. Although DBS can be remarkably effective, several challenges can limit its application and efficacy, including 1) the target neural structures are very small and challenging to accurately implant with electrodes; 2) DBS implantation surgery commonly requires the patient to be awake; 3) selection of beneficial stimulation parameters for chronic therapy is a laborious manual process prone to suboptimal outcomes; 4) chronic therapy is applied constantly regardless of the patient’s present therapeutic needs; and 5) the mechanisms of action of DBS are yet to be fully elucidated. Electrophysiological feedback signals may provide the means for addressing these challenges. A signal localised to the target neural structure could be used like a homing beacon to guide electrode implantation and configuration for chronic therapy. Furthermore, if the signal was measurable under general anaesthesia, it could be used to guide implantation in unconscious patients. A signal that varies with patient symptomatic state and the application of DBS could also guide selection of stimulation parameters, inform on mechanisms of action, and act as a feedback signal for monitoring patient state in real-time and automatically adapting stimulation settings to optimise therapy. This thesis identifies a novel evoked potential – termed evoked resonant neural activity (ERNA) – elicited by DBS pulses applied in the vicinity of the STN and investigates its potential for improving DBS therapy. ERNA is confirmed to be of neural origin and is shown to have largest amplitude in the dorsal subregion of the STN, where DBS for Parkinson’s disease is typically most effective. ERNA and its localisation are also shown to be present and readily measurable under general anaesthesia. Additionally, both the frequency and amplitude of ERNA are shown to vary with the application of therapeutically-effective DBS. The characteristics of ERNA make it an electrophysiological signal with considerable potential for addressing the challenges associated with applying DBS therapy, particularly STN-DBS. ERNA’s localisation to dorsal STN and presence under anaesthesia indicate utility in guiding electrode implantation and configuration in both awake and unconscious patients. Variation in ERNA with therapeutically-effective DBS also indicates potential utility in guiding the selection of chronic stimulation parameters, investigating mechanisms of action, and automatically adapting therapy according to the patient’s real-time needs.
-
ItemBehavioural and objective measurements of listening effort( 2021)Listening effort has been defined as the mental effort exerted to understand auditory stimuli (Pichora-Fuller et al, 2016a). In many cases it is one of the causes of isolation and poor quality of life for people with hearing aids and cochlear implants. One way to mitigate this problem is to assess listening effort during the fitting of hearing devices and then use these measurements to optimise device settings to improve audibility while at the same time minimising listening effort. However, at the moment, clinical practice lacks a standardised protocol for the assessment of listening effort. In this project, a behavioural test for the assessment of task difficulty was developed and validated. As was the feasibility of using fNIRS for the assessment of listening effort. The first study in this thesis focused on the development and validation of a behavioural dual task test for assessment of task difficulty. The test consisted of identification of the last word of each sentence in lists of seven sentences each, and recall of those last words after the presentation of each list. The test was implemented with Australian speech material, to minimise possible confounding factors due to accent and dialect dissonances when applied to the Australian population. The test was applied to twenty-four normal hearing participants. The results showed that, both identification and recall scores could predict differences in difficulty between speech presented in noise and speech presented in noise in which a binary mask noise reduction algorithm was applied to the speech mixture. The implemented test could be used in the Australian research practice to compare different hearing devices and processing algorithms at high levels of speech intelligibility. The second study investigated the feasibility of using fNIRS for assessment of listening effort. More specifically, in this study fNIRS was used for identifying cortical neuronal biomarkers of listening effort in the left inferior frontal gyrus (LIFG) and bilateral posterior superior temporal gyrus (pSTG). The findings of the study suggest the existence of task related concentration changes of oxy- and deoxy- haemoglobin during the expected exertion of listening effort. Furthermore, when listening effort was modulated by manipulation of motivation and task demands, the change in concentration of oxyhaemoglobin, followed the behaviour expected for listening effort. However, the results of this study were not statistically significant due to the high level of uncontrolled variability on the data. These results suggest a potential for fNIRS in the assessment of listening effort, however, more investigation needs to be done to optimise the protocols for assessment of listening effort and improve effect size. The third study continued with the investigation of the feasibility of using fNIRS for the assessment of listening effort. However, in this study, rather than using fNIRS for monitoring cortical neuronal activity, it was used for monitoring concentration changes of total haemoglobin in the extracerebral compartment (scalp and skull), associated with the activation of the autonomic nervous system. The findings of this study suggested the existence of task related concentration changes of total haemoglobin in the extracerebral compartment. When compared with the expected behaviour of listening effort modulated by manipulation of motivation and task demands, it was found that the concentration changes of total haemoglobin in the extracerebral compartment followed the expected trend of behaviour of listening effort, but again, these results were not statistically significant. The results suggests that fNIRS could potentially be used for assessment of listening effort, not just due to the capability of monitoring cortical neuronal activity, but also due to the capability of monitor activation in the autonomic nervous system. In this project, behavioural and objective measurements of listening effort were investigated. The work reported in this thesis has shown that behavioural methods can be used for assessment of listening task difficulty. But also, the experiments reported in this thesis investigated the feasibility of fNIRS for the assessment of listening effort. It was determined that fNIRS has the potential for assessment of listening effort because it facilitates the study of its different aspects, associated with cortical neural activity, but also, with activation of the autonomic nervous system when listening under challenging conditions. This work provides a base for future research that should be focused in determining a protocol for measurement of listening effort that integrates cortical neuronal and systemic components extracted from the fNIRS signal to better understand listening effort.
-
ItemIdentifying and addressing limitations in deep brain stimulation for Parkinson's disease using novel neuronal biomarkers( 2020)Deep brain stimulation (DBS) is an effective treatment in Parkinson’s disease (PD). DBS is postulated to modulate and restore ‘functionality’ to the pathological brain networks implicated in PD. DBS therapy improves motor symptoms and quality of life and allows for substantial medication reduction. However, there is little information on the long-term outcomes after DBS implantation including programming rates, battery changes, hardware complication surgeries or the duration of the therapy. This knowledge is crucial to inform on clinical decision making, distribution of healthcare resources and guide research direction for device development. The first part of this thesis presents a cross-sectional, population-based study of 1849 patients with PD implanted with DBS in Australia over a 15-year period (2002-2016). Individual-level patient data was derived from three linked national government databases and the requirements for DBS care and servicing was evaluated, referenced from the time of surgery. The mean annual programming rate was 6.9 in the first year, and 2.8 annually thereafter. Over 50% of patients required repeat hardware surgery after DBS implantation. 11.3% of patients had repeat intracranial electrode surgery (including 1.1% of patients who were completely explanted). 47.6% of patients had repeat implantable pulse generator/extension-cable surgery including for presumed battery depletion. 6.2% of patients had surgery of the implantable pulse generator /extension-cable surgery within one year of any previous such surgery. 30-day post-operative mortality was 0.3% after initial DBS implantation and 0.6% after any repeat hardware surgery. The median time from DBS surgery to residential care admission was 10.2 years and, to death was 11.4 years. These findings support development of technologies to reduce therapy burden such as enhanced surgical navigation, hardware miniaturisation and improved battery efficiency. The second part of the thesis focuses on two key contributors to DBS therapy burden, device programming and electrode redo surgery. DBS efficacy relies on the delivery of stimulation to the ideal location to achieve optimal motor benefit. This site is determined by the electrode location and the contact selected to apply DBS by the clinician. DBS programming is an arduous, time-consuming process, and highly dependent on clinician expertise. Neuronal signals have been proposed as biomarkers to assist in DBS programming and guide electrode implantation. The most widely studied signals are local field potentials (LFPs) such as beta oscillations and high frequency oscillations (HFO). However, LFPs hold recording challenges due to their small size and low signal to noise ratio. More recently, our group has described an evoked potential, elicited by DBS, termed ‘evoked resonant neural activity’ (ERNA). ERNA is a large amplitude signal, with a characteristic decaying oscillation morphology, that is reliably recordable in patients with PD implanted with subthalamic nucleus (STN) DBS. It localises to the postulated ideal anatomical location to apply DBS, in the dorsal STN. However, the clinical relevance of ERNA and its utility compared to LFPs or electrode anatomical location is unknown. Thus, in 50 (100 hemispheres) consecutive patients with PD implanted with STN DBS, ERNA power, beta power and HFO power was measured from each of the four contacts on the DBS lead during surgery. Neuroimaging was obtained peri-operatively to visualise each contact and determine its proximity to a nominated ideal anatomical location. The four contacts in each hemisphere were ranked from one to four according to neuronal signal power and anatomical location. In 14 patients (28 hemispheres), therapeutic stimulation was applied to each of the four contacts on the DBS lead and the degree of motor benefit measured. ERNA, beta oscillations and the anatomical location of contacts similarly predicted how motor benefit varied across contacts with STN DBS therapy. Combining ERNA, beta and HFO ranking data yielded the strongest predictive model. However, only first-ranked contacts according to ERNA delivered a motor benefit that was significantly better than at the other three contacts on the lead. Furthermore, only first-ranked ERNA contacts delivered a motor benefit that was not significantly less than the maximal available in each hemisphere. When monopolar configuration was chosen at 6 months after DBS surgery, the clinician-chosen contact corresponded most frequently with the first-ranked ERNA contact in 81% of hemispheres compared to 71% for anatomy and 52% for beta. ERNA performed significantly better than anatomy and beta oscillations and combining data using machine learning algorithms did not improve model performance compared to using ERNA data alone. These results support the role of ERNA in guiding contact selection and assisting intra-operative electrode navigation to ultimately, reduce the treatment burden of DBS therapy.
-
ItemOculomotor Behaviour and Perceptual Localisation in Retinal Prostheses( 2020)Prosthetic vision is an emerging technology aiming to provide artificial vision to the profoundly blind. Present-day visual prostheses provide useful assistance in everyday life but the quality of vision is poor, with only fractional visual field coverage and limited resolution of detail. The retina, optic nerve, or primary visual cortex are electrically stimulated using implanted electrodes to elicit artificial visual percepts. Eye movements cause movement of the percept within the visual field, however, in many devices, electrode activity is modulated by images captured by a head-mounted camera that does not move in conjugate with eye movement. Percept locations are therefore dissociated from the real-world, leading to localisation errors. Users must be trained to use head movement, rather than eye movement, to control the camera, potentially increasing the difficulty of using the device for every-day activities. This thesis explores oculomotor behaviour in prosthetic vision and investigates the use of eye tracker feedback to redirect the video input in real time (‘gaze compensation’) to restore naturalistic control of gaze. The first study presented in the thesis investigated the effect of visual field loss on eye and head movement coordination in low-vision subjects with retinitis pigmentosa (RP), the current primary indication for retinal prostheses. Visual field loss was found to be associated with a habitually confined range of eye movement and a greater reliance on head movement. This has implications for training and rehabilitation in visual prostheses, as recipients with RP express atypical eye and head scanning behaviour. An investigation of the oculomotor behaviour of retinal prosthesis recipients in a forced-choice localisation task and a motion discrimination task is also presented. Although the participants were aware of the potential for eye movements to impair task performance, systematic eye movements were observed in response to the task stimuli. These were interpreted as reflexive eye movements made in response to the static and dynamic stimuli, as would be expected in normal vision, suggesting preserved oculomotor capacity. This is a promising indication for the naturalistic integration of artificially evoked percepts into the visual system, but the primary purpose of these eye movements, namely foveation, cannot be fulfilled without gaze compensation. Following the demonstration of naturalistic eye movement in retinal prosthesis users and the finding that suppression of eye movement was difficult or impossible, the thesis then examines the effect of eye movement on localisation and the possible benefits of gaze compensation. It was found that eye movement lead to localisation errors in a target localisation task in simulated prosthetic vision, but the introduction of gaze compensation resolved this. A subsequent pilot study in retinal prosthesis recipients is also presented, in which no benefit of gaze compensation for localisation was observed, possibly because the subjects had learned compensatory strategies. However, some methodological problems were identified, and similar studies from a different group do show a benefit of gaze compensation. Overall the thesis advances the understanding of the perceptual experience and oculomotor behaviour of visual prosthesis users and argues for the integration of gaze compensation in camera-based visual prostheses.