This research addresses challenges in two major applications, both related to photonic cognitive sensing. The first part, “Implantable Photonic Nano-Probe Detectors for Neural Imaging”, focuses on imaging system in the neural sciences field. The second part, “Advanced Control System for Optical Data Communications”, covers embedded low power control systems for optical communications. Implantable Photonic Nano-Probe Detectors for Neural Imaging This first part address the problem of simultaneous and real-time monitoring of dense brain neural activity, with the capability of cellular resolution and cell-type specificity included. For decades, electrophysiology has been the “gold standard” for the recording of neural activity. Despite recent advances, electrophysiology techniques can typically monitor fewer than 100 neurons simultaneously, due to the practical limits of electrode density. Additionally, the ability of direct monitoring specific cell types is not possible here. With the introduction of a growing panel of fluorescent optical reporters for brain function mapping, optical microscopy techniques have demonstrated the ability to track the activity of hundreds of neurons simultaneously in a much less invasive manner but with high spatial resolution, low-to-moderate temporal resolution and cell-type specificity. Unfortunately, only superficial layers of the brain can be imaged by free-space microscopy, due to the intrinsic light scattering and absorption limitation in brain tissue. To allow optical fluorescence imaging of deeper layers of the brain with proper a signal-to-noise ratio, a dense and scalable 3-D lattice of photo emitter and detector pixels (E-Pixels and D-Pixels, respectively) must be distributed on shanks for possible insertion into the brain. The 3-D lattice (combined fluorescent optical reporters) is expected to give an activity image of a very large neural population at an arbitrary depth in the brain. This work presents the design and implementation of the aforementioned 3-D photo- detectors (D-Pixels), associated with data processing and readout circuitries, for the future assembly of a probe-based system for functional imaging of neural activity. One of the main challenges of producing a probed-based version of a fluorescence microscope is the rejection of the light used to excite the fluorescent reporters. This is commonly done in the spectral domain with band-pass filters for free-space microscopy. However, these filters are not implementable with the proper optical density at the probe scale. The probe-based photo-detectors must be capable of rejecting the excitation light and capturing only the fluorescent response without the use of optical filters. Integrated Geiger-mode single-photon avalanche diodes (SPADs) are used as the sensing devices, which provide the ability to capture low fluorescence signals, fast response in the time domain, and direct digital readout. By engineering narrow E-Pixels angular-excitation fields and overlapping them with the narrow D-Pixels detection fields, fluorescent sources can be spatially localized. The detectors are embedded into four ultra-thin implantable shanks, associated with data processing units and readout circuits, all forming the photonic nano-probe detectors (also referred to as “D-Pixels Camera Chip (DCC)”). The shanks have dimensions of 110um×50um each, with 100 pixels along a shank (a total number of 400 pixels), distributed over 3mm length. The data generated by the photonic nano-probe detectors, is serially streamed out at a rate of 640Mbps, for offline analysis and image reconstruction. The photonic nano-probe detectors are fabricated in a conventional CMOS 0.13um technology. This part of the thesis first proposes and develops the architecture of the photonic nano-probe detectors. The challenges of designing high density, ultra-thin probes with the aforementioned form factor, fabricated in CMOS 0.13um technology is also discussed. Secondly, the design and implementation of testability and debugging options are mentioned, as playing an important role in achieving research goals. Last the design of lab experimental setups is presented and as well as the measurement results of the photonic nano-probe detectors. Experimental results indicate on achieving the crucial key features of the research work, the capability of rejecting the excitation light and capturing only the fluorescent decay response without the use of optical filters. Additionally, the results show that the photonic nano-probe detectors can precisely localize and map into a 2-D image, a light source within a pixel resolution.   Advanced Control System for Optical Data Communications The second part of the thesis focuses on the problem of initialization and temperature stabilization of silicon photonic (SiP) devices, with focus on dramatic power reduction of the power consumption. While microelectronics technology continues growing in scale, bandwidth, and integration of multiple systems on a single silicon die, the traditional electrical interconnects become the speed bottleneck in high-performance data communication systems. On the other hand, silicon photonics offers a promising platform for integration and manufacturing of photonics devices for high speed data transfer applications, such as access networks, supercomputers, chip-to-chip interconnects, and data centers. Additionally, the high index contrast of silicon platform and its compatibility with CMOS technologies, gives rise to integration of high speed, power efficient silicon photonic interconnects and most innovative CMOS technologies. Micro-ring resonators (MMRs), which are important building blocks is many silicon photonics applications, became attractive devices in many optical communication systems. This is due to their wavelength tuning ability, low power consumption and small footprint. However, temperature changes in their environment will shift their resonance from the desired point (due to high thermo-optical coupling in silicon), leading to performance degradation of the optical link. Compensating the degradation in performance can be directly translated to an excess in overall power consumption of the link, which will be critical in high-speed optical data communication systems. This work develops and demonstrates an ultra-low power control system, for initialization and temperature stabilization of MMRs. It utilizes an integrated heater, to thermally tune and lock the resonator to the desired wavelength. Traditional feedback loops rely on tapping a portion of the optical signal with the use of integrated photodiodes. They lock on the desired wavelength by sensing the maximum signal intensity, observed by the photodiode. The suggested control system in this work is based on an analog control system and utilizes the photo-conductance effect of doped-resistive heaters, to sense the optical power through the micro-ring. This part of the thesis first develops a VERILOG-A model for the photo-conductance effect of the doped-resistive heater. This enables the integration of the heater’s model with the proposed control circuits, into a circuit design simulator. Secondly, an architecture for the control system is proposed and developed, which includes fundamental electronic circuits with the aforementioned heater’s model. For the purpose of circuit level simulations, a design methodology is developed, which is based on semi-ideal models for the electronic building blocks. Then a circuit level simulator is used to simulate and evaluate the performance of the control system. Last, the proposed system is implemented with the use of commercial discrete electronic components, all connected on a custom designed printed circuit board (PCB). Simulations of the control system indicate an initialization time less than 160us, and maximum locking voltage error of 1.8%. The obtained dynamic energy consumption is ED=85 fJ/bit/oC for bit rate of 20Gbps. Though the control system is targeted for MRRs, it can be easily expanded to control other PIC devices.

One of the central questions in neuroscience is how the nervous system generates and regulates behavior. To understand the neural code for any behavior, an ideal experiment would entail (i) quantitatively defining that behavior, (ii) recording neuronal activity in relevant brain regions to identify the underlying neuronal circuits and eventually (iii) manipulating them to test their function. Novel methods in neuroscience have greatly advanced our abilities to conduct such experiments but are still insufficient. In this thesis, I developed methods for these three goals. In Chapter 2, I describe an automatic behavior identification and classification method for the cnidarian Hydra vulgaris using machine learning. In Chapter 3, I describe a fast volumetric two-photon microscope with dual-color laser excitation that can image in 3D the activity of populations of neurons from visual cortex of awake mice. In Chapter 4, I present a machine learning method that identifies cortical ensembles and pattern completion neurons in mouse visual cortex, using two-photon calcium imaging data. These methods advance current technologies, providing opportunities for new discoveries.

This thesis contributes to recent interest within medical imaging regarding the development and clinical application of magnetic resonance elastography (MRE) to the human brain. MRE is a non-invasive phase-contrast MRI technique for measurement of brain mechanical properties in vivo, shown to reflect the composition and organisation of the complex tissue microstructure. MRE is a promising imaging biomarker for the early characterisation of neurodegeneration due to its exquisite sensitivity to variation among healthy and pathological tissue. Neurodegenerative diseases are debilitating conditions of the human nervous system for which there is currently no cure. Novel biomarkers are required to improve early detection, differential diagnosis and monitoring of disease progression, and could also ultimately improve our understanding of the pathophysiological mechanisms underlying degenerative processes. This thesis begins with a theoretical background of brain MRE and a description of the experimental considerations. A systematic review of the literature is then performed to summarise brain MRE quantitative measurements in healthy participants and to determine the success of MRE to characterise neurological disorders. This review further identified the most promising acquisition and analysis methods within the field. As such, subsequent visits to three brain MRE research centres, within the USA and Germany, enabled the acquisition of exemplar phantom and brain data to assist in discussions to refine an experimental protocol for installation at the Edinburgh Imaging Facility, QMRI (EIF-QMRI). Through collaborations with world-leading brain MRE centres, two high-resolution - yet fundamentally different - MRE pipelines were installed at the EIF-QMRI. Several optimisations were implemented to improve MRE image quality, while the clinical utility of MRE was enhanced by the novel development of a Graphical User Interface (GUI) for the optimised and automatic MRE-toanatomical coregistration and generation of MRE derived output measures. The first experimental study was performed in 6 young and 6 older healthy adults to compare the results from the two MRE pipelines to investigate test-retest agreement of the whole brain and a brain structure of interest: the hippocampal formation. The MRE protocol shown to possess superior reproducibility was subsequently applied in a second experimental study of 12 young and 12 older cognitively healthy adults. Results include finding that the MRE imaging procedure is very well tolerated across the recruited population. Novel findings include significantly softer brains in older adults both across the global cerebrum and in the majority of subcortical grey matter structures including the pallidum, putamen, caudate, and thalamus. Changes in tissue stiffness likely reflect an alteration to the strength in the composition of the tissue network. All MRE effects persist after correcting for brain structure volume suggesting changes in volume alone were not reflective of the detected MRE age differences. Interestingly, no age-related differences to tissue stiffness were found for the amygdala or hippocampus. As for brain viscosity, no group differences were detected for either the brain globally or subcortical structures, suggesting a preservation of the organisation of the tissue network in older age. The third experiment performed in this thesis finds a direct structure-function relationship in older adults between hippocampal viscosity and episodic memory as measured with verbal-paired recall. The source of this association was located to the left hippocampus, thus complementing previous literature suggesting unilateral hippocampal specialisation. Additionally, a more significant relationship was found between left hippocampal viscosity and memory after a new procedure was developed to remove voxels containing cerebrospinal fluid from the MRE analysis. Collectively, these results support the transition of brain MRE into a clinically useful neuroimaging modality that could, in particular, be used in the early characterisation of memory specific disorders such as amnestic Mild Cognitive Impairment and Alzheimer's disease.

Large-scale neuroimaging data is becoming increasingly available, providing a rich data source with which to study neurological conditions. In this thesis, I demonstrate the utility of large-scale neuroimaging as it applies to Alzheimer’s disease (AD) and normal aging, using univariate parametric mapping, regional analysis, and advanced machine learning. Specifically, this thesis covers: 1) validation and extension of prior studies using large-scale datasets; 2) AD diagnosis and normal aging evaluation empowered by large-scale datasets and advanced deep learning algorithms; 3) enhancement of cerebral blood volume (CBV) fMRI utility with retrospective CBV-fMRI technique. First, I demonstrated the utility of large-scale datasets for validating and extending prior studies using univariate analytics. I presented a study localizing AD-vulnerable regions more reliably and with better anatomical resolution using data from more than 350 subjects. Following a similar approach, I investigated the structural characteristics of healthy APOE ε4 homozygous subjects screened from a large-scale community-based study. To study the neuroimaging signatures of normal aging, we performed a large-scale joint CBV-fMRI and structural MRI study covering age 20-70s, and a structural MRI study of normal aging covering the full age-span, with the elder group screened from a large-scale clinic-based study ensuring no evidence of AD using both longitudinal follow-up and cerebrospinal fluid (CSF) biomarkers evidences. Second, I performed deep learning neuroimaging studies for AD diagnosis and normal aging evaluation, and investigated the regionality associated with each task. I developed an AD diagnosis method using a 3D convolutional neural network model trained and evaluated on ~4,600 structural MRI scans and further investigated a series of novel regionality analyses. I further extensively studied the utility of the structural MRI summary measure derived from the deep learning model in prodromal AD detection. This study constitutes a general analytic framework, which was followed to evaluate normal aging by performing deep learning-based age estimation in cognitively normal population using more than 6,000 scans. The deep learning neuroimaging models classified AD and estimated age with high accuracy, and also revealed regional patterns conforming to neuropathophysiology. The deep learning derived MRI measure demonstrated potential clinical utility, outperforming other AD pathology measures and biomarkers. In addition, I explored the utility of deep learning on positron emission tomography (PET) data for AD diagnosis and regionality analyses, further demonstrating the broad utility and generalizability of the method. Finally, I introduced a technique enabling CBV generation retrospectively from clinical contrast-enhanced scans. The derivation of meaningful functional measures from such clinical scans is only possible through calibration to a reference, which was built from the largest collection of research CBV-fMRI scans from our lab. This method was validated in an epilepsy study and demonstrated the potential to enhance the utility of CBV-fMRI by enriching the CBV-fMRI dataset. This technique is also applicable to AD and normal aging studies, and potentially enables deep learning based analytic approaches applied on CBV-fMRI with similar pipelines used in structural MRI. Collectively, this thesis demonstrates how mechanistic and diagnostic information on brain disorders can be extracted from large-scale neuroimaging data, using both classical statistical methods and advanced machine learning.

Segmentational grouping in music listening refers to the organisation of individual tones into tone groups that tend to be processed and subsequently recalled as perceptual units or chunks. Grouping of tones via this process tends to occur at natural breaks in structure of a tone sequence, so that relatively larger changes in pitch, amplitude or timing are perceived as boundaries which cue the segmentational grouping process. Segmentational grouping processes have been examined using behavioural research techniques; yet neurophysiological processes underlying the grouping process have received little attention, and are poorly understood. This study aimed to identify brain regions involved in the segmentational grouping process as cued by rhythmic information. Participants performed two auditory tasks while brain electrical activity responses were monitored using Steady-State Probe Topography (SSPT). Behavioural responses evoked in a task probing individuals' use of lengthened-duration tones to organise memory for pitch sequences indicated that longer-duration tones were used as cues to organise working memory representations of the musical patterns. Examination of dynamic SSPT responses during the encoding phase of a probe recognition task indicated that greater use of rhythmic cues to organise working memory representations was associated with activation of a network of left hemisphere frontal, temporal and parietal regions. During the lengthened tone, activation of left central and vertex regions and progressive activation of left temporal and temporoparietal regions were linked with use of the deviant status of the lengthened tone to update temporal expectations for the sequence. Excitatory responses observed in left posterior frontal and temporal regions to a tone following the lengthened tone were proposed to reflect temporal allocation of attention to this point in time, whereas sustained excitatory activation of left temporal, and temporoparietal regions reflected the role of these regions in supporting representations of the tone events in working memory. Finally, late inhibitory responses to the tone following the lengthened tone in left frontal, temporal, temporoparietal, and parietal regions were linked with the manipulation and closure of the working memory trace in association with the grouping process. Together, these findings support the activation of a network of left frontal, temporal and parietal regions underlying rhythmic grouping of pitch sequences.

This study investigated the neural bases underlying representation of nouns and verbs at the semantic, lexical form, and morpho-syntactic levels in Mandarin Chinese, a language with little inflectional morphology. Compared with other studies employing European languages with rich inflections, examination of Chinese would allow the separation of conceptual and morpho-syntactic operations based on different stimulus formats and experimental paradigms. To deal with both the theoretical and design issues in previous studies, several additional measures were taken. First, at each cognitive level, two experiments, one receptive and one expressive, were conducted. Moreover, convergence across experiments at the same cognitive level was computed in order to search for taskindependent grammatical class effects. Second, both concrete and abstract nouns and verbs were included, and conjunction analyses across the two concreteness levels were employed to ensure the generalizability of the findings to all nouns and verbs. Results revealed greater activation for verbs in the left posterior lateral temporal gyri in experiments at both semantic and morpho-syntactic levels, and stronger responses in the prefrontal cortex, including left BA47 and the supplementary motor area, only for morpho-syntactic processing associated with nominal grammatical morphemes, namely, classifiers. No differential levels of activation for nouns and verbs were observed in tasks emphasizing word form representation. While greater activation for processing of nominal classifiers in prefrontal areas may reflect differences in computational complexity associated with selection of grammatical morphemes, the involvement of left posterior lateral temporal cortex has been interpreted as reflecting semantic processing of verbs. The nature of processes represented in each of these regions was further discussed with findings from previous relevant studies. Finally, future studies are proposed for further exploration into the neural mechanisms underlying presentation of nouns and verbs using more recently developed methods of analyses. / published_or_final_version / Speech and Hearing Sciences / Doctoral / Doctor of Philosophy

No description available.

Individuals with mild cognitive impairment (MCI) show variable impairment in autobiographical memory function, source memory function and reduced integrity in the brain’s default mode network (DMN). There is overlap between the DMN, such as the medial posterior cortical hub, and brain regions that are active when participants recall autobiographical memories. To assess the association between autobiographical memory and the DMN, 14 MCI and eleven age and education-matched healthy control participants were assessed using the autobiographical memory interview (AMI) and underwent resting state fMRI scans. The same participants underwent a test of source memory which assessed both recognition and source memory. The MCI group showed significantly increased semantic as well episodic memory impairments using the AMI, evident across the lifespan for episodic memory but not for childhood semantic memory. Significantly poorer DMN connectivity, using a goodness of fit index (GOF) of the DMN template, was evident in the MCI group. MCI participants showed poorer performance on both recognition and source memory relative to HC participants. A modest association between AMI semantic memory (r=0.4) scores, but not episodic memory scores (r=0.09), and DMN connectivity was found in these participants. For future study the predictive value of MR imaging in the DMN of MCI participants should be explored.

Thesis (M.S.) -- Worcester Polytechnic Institute. / Keywords: algorithm; segmentation; rat atlas; MRI. Includes bibliographical references (p. 77-82).

Matrix factorisation treats observations as linear combinations of basis vectors together with, possibly, additive noise. Notable techniques in this family are Principal Components Analysis and Independent Components Analysis. Applied to brain images, matrix factorisation provides insight into the spatial and temporal structure of data. We improve on current practice with methods that unify different stages of analysis simultaneously for all subjects in a dataset, including dimension estimation and reduction. This results in uncertainty information being carried coherently through the analysis. A computationally efficient approach to correlated multivariate normal distributions is set out. This enables spatial smoothing during the inference of basis vectors, to a level determined by the data. Applied to neuroimaging, this reduces the need for blurring of the data during preprocessing. Orthogonality constraints on the basis are relaxed, allowing for overlapping ‘networks’ of activity. We consider a nonparametric matrix factorisation model inferred using Markov Chain Monte Carlo (MCMC). This approach incorporates dimensionality estimation into the infer- ence process. Novel parallelisation strategies for MCMC on repeated graphs are provided to expedite inference. In simulations, modelling correlation structure is seen to improve source separation where latent basis vectors are not orthogonal. The Cambridge Centre for Ageing and Neuroscience (Cam-CAN) project obtained fMRI data while subjects watched a short film, on 30 of whose recordings we demonstrate the approach. To conduct inference on larger datasets, we provide a fixed dimension Structured Matrix Factorisation (SMF) model, inferred through Variational Bayes (VB). By modelling the components as a mixture, more general distributions can be expressed. The VB approach scaled to 600 subjects from Cam-CAN, enabling a comparison to, and validation of, the main findings of an earlier analysis; notably that subjects’ responses to movie watching became less synchronised with age. We discuss differences in results obtained under the MCMC and VB inferred models.