Engineering Science - Theses, Dissertations, and other Required Graduate Degree Essays

Receive updates for this collection

Use of functional correlation tensors for fMRI monitoring of neuroplasticity during motor learning

Author: 
Date created: 
2020-05-11
Abstract: 

The development of Functional Correlation Tensors (FCT) is driving novel investigations into whole-brain functional magnetic resonance (fMRI) signal synchronicity. FCTs are mathematically analogous to the established structural diffusion modality. Unlike conventional fMRI analysis, FCTs examine functional signal independently of hemodynamic response assumptions. In this work participants trained on a motor task for two weeks with fMRI and diffusion scans collected at baseline and endpoint. Using only baseline data, a significant correlation was detected for the fractional anisotropy of the diffusion data with the signal synchronicity anisotropy of the FCT data. Previous work on this data detected white matter (WM) neuroplasticity in motor regions between baseline and endpoint. As FCT is sensitive to WM function, it was hypothesized that WM neuroplasticity could be further detected. Significant increases in signal synchronicity were detected in areas of motor task planning and execution. This represents the first instance of this novel methodology for identifying neuroplasticity.

Document type: 
Thesis
File(s): 
Senior supervisor: 
Ryan D'Arcy
Department: 
Applied Sciences: School of Engineering Science
Thesis type: 
(Thesis) M.A.Sc.

3D convolutional neural networks for Alzheimer’s disease classification

Author: 
Date created: 
2020-01-21
Abstract: 

Dementia of the Alzheimer’s type (DAT) is a neurodegenerative disease characterized by abnormal brain metabolism and structural brain atrophy. These functional and structural changes can be observed in images acquired using 18F-fluorodeoxyglucose positron emission tomography (FDG-PET) and structural magnetic resonance imaging (MRI). Traditional machine learning framework for DAT classification often involves time-consuming segmentation of brain images as part of the feature extraction process, while deep neural networks can learn DAT-related patterns directly from brain images to generate DAT probability scores. In this thesis, we design 3D convolutional neural networks (CNN) for two applications: classification and segmentation. We design classification networks for single modality use and perform comprehensive evaluation by measuring the performance of our networks on images along the entire DAT spectrum. To support traditional DAT classification framework, we design a fast and accurate segmentation pipeline. We propose a hemisphere-based approach where we train networks to localize and segment hemispheres.

Document type: 
Thesis
File(s): 
Senior supervisor: 
Faisal Beg
Department: 
Applied Sciences: School of Engineering Science
Thesis type: 
(Thesis) M.A.Sc.

Magnetic composite polymer membrane actuators with applications to microfluidic devices

Author: 
Date created: 
2020-04-08
Abstract: 

This thesis demonstrates new materials and microfabrication techniques for integrating membrane-type magnetic composite polymer (M-CP) actuators into microfluidic devices and systems. A membrane actuator with a powerful stroke volume that displaces 7.4 µL of water under a 110mT external magnetic field is developed and demonstrated in a hybrid M-CP/thermoplastic microfluidic device and in an all-PDMS microfluidic device. To achieve injection mouldable M-CP devices, a new M-CP is developed that consists of an injection mouldable off-thiol-ene-epxoy (OSTE+) polymer resin that is embedded with 25 weight-% rare earth magnet particles to be permanently magnetized. To support the rapid prototyping of PDMS and OSTE+ polymer microfluidic devices, a new type of micromould is developed that uses laser ablation of tape to deliver low cost, ultra-rapid moulds. These developments facilitate future commercial mass production through integration with thermoplastic polymers favored by the microfluidics industry in a scalable, “design-to-manufacture” scheme.

Document type: 
Thesis
File(s): 
Senior supervisor: 
Bonnie Gray
Department: 
Applied Sciences: School of Engineering Science
Thesis type: 
(Thesis) M.A.Sc.

Tensor completion methods for collaborative intelligence

Author: 
Date created: 
2020-03-20
Abstract: 

In the race to bring Artificial Intelligence (AI) to the edge, collaborative intelligence has emerged as a promising way to lighten the computation load on edge devices that run applications based on Deep Neural Networks (DNNs). Typically, a deep model is split at a given layer into edge and cloud sub-models. The deep feature tensor produced by the edge sub-model is transmitted to the cloud, where the remaining computationally intensive workload is performed by the cloud sub-model. The communication channel between the edge and cloud is imperfect, which will result in missing data in the deep feature tensor received at the cloud side, an issue that has mostly been ignored by existing literature on the topic. In this thesis I study four methods for recovering missing data in the deep feature tensor. Three of the studied methods are existing, generic tensor completion methods, and are adapted here to recover deep feature tensor data, while the fourth method is newly developed specifically for deep feature tensor completion. Simulation studies show that the new method is 3 − 18 times faster than the other three methods, which is an important consideration in collaborative intelligence. For VGG16’s sparse tensors, all methods produce statistically equivalent classification results across all loss levels tested. For ResNet34’s non-sparse tensors, the new method offers statistically better classification accuracy (by 0.25% − 6.30%) compared to other methods for matched execution speeds, and second-best accuracy among the four methods when they are allowed to run until convergence.

Document type: 
Thesis
File(s): 
Senior supervisor: 
Ivan Bajic
Department: 
Applied Sciences: School of Engineering Science
Thesis type: 
(Thesis) M.A.Sc.

Visually-guided beamforming for a circular microphone array

Author: 
Date created: 
2020-04-06
Abstract: 

Beamforming is a technique which can adaptively steer the pattern of a microphone array towards or away from a target direction. Three conventional beamforming techniques are reviewed and compared with a beamformer proposed here, called MVDR-2C. Most acoustic beamformers selectively locate a single desired sound source, such as a speaker, and the beamforming performance drops significantly when two or more speakers are active. In order to deploy beamforming in a room, a circular microphone array is supplemented by a 360° camera comprising two fisheye lenses. The camera allows face detection to provide the speaker directions to the beamformer. In order to develop face/object detectors that operate directly on fisheye images, three annotated fisheye datasets are generated and used to re-train an existing face detector. Finally, several beamformers are evaluated and compared, demonstrating the clear performance advantage of the proposed one.

Document type: 
Thesis
File(s): 
Senior supervisor: 
Ivan Bajic
Department: 
Applied Sciences: School of Engineering Science
Thesis type: 
(Thesis) M.A.Sc.

Pseudo-real-time Retinal Layer Segmentation for OCT

Date created: 
2019-12-19
Abstract: 

In this thesis, we present a pseudo-real-time retinal layer segmentation for high-resolution Sensorless Adaptive Optics-Optical Coherence Tomography (SAO-OCT). Our pseudo-real-time segmentation method is based on Dijkstra’s algorithm that uses the intensity of pixels and the vertical gradient of the image to find the minimum cost in a geometric graph formulation in a limited search region. It segments six retinal layer boundaries in an iterative process according to their order of prominence. The segmentation time is strongly related to the number of retinal layers to be segmented. Our program permits en face images to be extracted during data acquisition to guide the depth specific focus control and depth dependent aberration correction for high-resolution SAO-OCT systems. The average processing times for our entire pipeline for segmenting six layers in a retinal B-scan of 992x400 pixels, 496x400 pixels and 240x400 pixels are around 23 ms, 26 ms and 14 ms, respectively.

Document type: 
Thesis
File(s): 
Senior supervisor: 
Marinko V. Sarunic
Yifan Jian
Department: 
Applied Sciences: School of Engineering Science
Thesis type: 
(Thesis) M.A.Sc.

Exploratory study to use K-means clustering for gesture selection of force myography upper limb data in participants with cerebral palsy

Author: 
Date created: 
2019-12-04
Abstract: 

Many with Cerebral Palsy (CP) use assistive devices to perform daily activities. A gesture recognition based wearable device can be implemented using force myography (FMG). However, little research has been done regarding gestures to use with populations that exhibit physical disturbances associated with CP. The research presented in this Thesis lays the groundwork for implementing k-means clustering to conduct gesture selection for a FMG wearable device in a clinical setting. The concept was tested with ten healthy participants and then validated in a pilot study with a CP participant. The results from both population studies showed that the k-means clustering is able to determine the ideal gesture subset in a shorter computation time than testing machine learning models with all the possible combinations of gestures. A finally study explored online testing with three healthy participants controlling a line-following robot with the FMG band. Though this work provides the foundation for using the FMG technology to interact with individuals with cerebral palsy, additional studies are required to determine its full potential.

Document type: 
Thesis
File(s): 
Senior supervisor: 
Carlo Menon
Department: 
Applied Sciences: School of Engineering Science
Thesis type: 
(Thesis) M.A.Sc.

Antenna designs for modern applications and smart radar for fall detection

Date created: 
2019-12-12
Abstract: 

The antenna is the most critical single component of a communication and sensing system. It not only provides the basic transformation between the electrical signals and electromagnetic waves, but also governs the signal-to-noise ratio which limits the performance of the signal processing and the entire system. The antenna design - both construction and pattern - must account for different system requirements. In diversity/MIMO applications, where the goal is to mitigate multipath-induced signal fading (or improve capacity, error rate, range, coverage, and a host of other performance metrics), it is normally beneficial to have broad patterns for matching to the broad angular range of the multipath, and polarization purity is not a priority. In a polarimetric radar, the target is usually within a narrow angular range, and different combinations of polarizations in the transmit and receive provide different information. These connections between the antenna and the signal processing, for different applications, motivate the new designs presented in this dissertation. The first part concerns multi-element designs for diversity and MIMO, used for portable terminals in broadcast, Wi-Fi and cellular systems. Performance evaluation using the patterns and statistical models for the multipath propagation is the key design tool. The von-Mises Fisher distribution is introduced for evaluating the impact of directivity in the multipath. The antenna construction is typically PCB-based since the products must be very low cost and compact. The second part strives for higher directivity using new designs of fixed arrays. These designs include dual-polarization, multiple frequency bands, and circular polarization. The construction is slotted metallic cavities because of the low loss in both the elements and the feed (distribution of power over the aperture), and the potential simplicity of manufacture, given the higher directivity of polarized illumination. The final part discusses new radar signal processing for indoor fall detection. A radar system was developed and tested, and demonstrates the detection of falls, breathing, and other movements, even when a person has fallen and is on the floor. Deep learning algorithms are used with new experiments providing the training data for distinguishing a person from other moving entities such as pets, reducing the false-alarm rate of the fall detection.

Document type: 
Thesis
File(s): 
Senior supervisor: 
Rodney Vaughan
Department: 
Applied Sciences: School of Engineering Science
Thesis type: 
(Thesis) Ph.D.

Design of a wearable to improve single-device motion classification of arm motions

Author: 
Date created: 
2019-11-26
Abstract: 

Inertial Measurement Unit (IMU) based wearable sensors have found common use to track arm activity in daily life. However, classifying a high number of arm motions with single IMU-based systems remains a challenging task. In this study, we propose a single-device wearable which incorporates a thermal sensor and an inertial sensor. The system was evaluated in a study incorporating 11 healthy participants, where 24 different arm motions were recorded and predicted with a machine learning classifier. This study found that 24 arm motions could be classified with 93.55% accuracy. Further, the passive infrared thermal sensor significantly increased classification accuracy from 75% to 93.55% , p=<0.05. The performance of the generalized classifier indicates that the device could classify arm motions on a user without prior training.

Document type: 
Thesis
File(s): 
Senior supervisor: 
Carlo Menon
Department: 
Applied Sciences: School of Engineering Science
Thesis type: 
(Thesis) M.A.Sc.

Brain vital signs: Towards next generation neurotechnologies for rapid brain function assessments at point-of-care

Author: 
Date created: 
2019-11-18
Abstract: 

Vital signs such as heart rate, blood pressure and body temperature have revolutionized medical care by providing rapidly assessed, physiology-based, non-invasive and easy-to-understand standardized metrics of different body functions. However, no such vital sign exists for the brain; instead, assessments of the brain are largely reliant on surrogate measures such as observations of behaviour or questionnaire-based measurements, which have been shown to be subjective and unreliable. This research aims to fill this key scientific, clinical, and technological gap by developing a brainwave-based technology platform to evaluate ‘vital sign’ metrics for the brain. A series of studies were undertaken to create and demonstrate a ‘brain vital signs’ platform that is capable of assessing a broad spectrum of functions ranging from the lower-level functions (i.e. sensation) to the highest-level cognition domains (i.e. contextual orientation). In particular, the first study focused on development and initial demonstration of the methods and apparatus for the brain vital signs technology; the next study focused on characterizing the brain vital sign responses to ensure scientific validity; the third study focused on creating a previously non-existant neurophysiology-based neural marker capable of capturing contextual orientation – which is the highest level cognitive domain known to be crucial to frontline clinical assessments; and finally, the last study focused on developing an advanced data analytic technique for maximizing signal capture under noisy environments typical of point-of-care evaluation settings. This research represents the first time that a ‘vital sign’-like metric has been developed for the brain that embodies the key characteristics of existing vital signs, enabling brain function measures that are rapid (~5 minute testing time), easy to use, portable, non-invasive, and standardized with automated analysis. Crucially, these vital sign metrics directly measure the brain’s electrical activity and do not depend on any responses from the test participant, thus providing much more objective information about brain function. The development of portable and objective ‘vital sign’-like metrics for the brain not only advances the scientific understanding of brain function through novel metrics like orientation, but also creates significant opportunities for enhancing clinical diagnosis through improved brain function assessments at the point-of-care.

Document type: 
Thesis
File(s): 
Senior supervisor: 
Ryan D'Arcy
Department: 
Applied Sciences: School of Engineering Science
Thesis type: 
(Thesis) Ph.D.