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

Receive updates for this collection

Asymmetric Coherent Configurable Caches for PolyBlaze Multicore Processor

Author: 
Date created: 
2014-11-28
Abstract: 

Modern computing systems gain performance by several means such as increased parallelism through using Chip-level Multiprocessor (CMP) systems. Symmetric Multiprocessor (SMP) systems use uniform processing cores to form a CMP in which all cores are identical in every aspect. Conversely, Asymmetric Multiprocessor (AMP) systems consist of processing cores with variable configurations such as different cache configurations, co-processors, and cache sizes. AMP systems coupled with such smart scheduling algorithms can improve resource utilization while maintaining overall system performance because real-time profiling in a computing system using light-weight hardware profilers can help smart scheduling algorithms make meaningful decisions. In other words, the vision into an application’s behavior helps in the decision making process on how to allocate available resources for different applications without penalizing the performance by putting too much overhead on the system. Currently, there is no AMP research framework available that allows us to look into asymmetry in processing systems. In this thesis, we present an extension on PolyBlaze framework for asymmetric coherent Level-1 (L1) caches. Our implementation in this work includes other arbiter and prefetching units as well. We measure data cache read miss rates and application run-times for select benchmarks from SPEC CPU2006 executed in a Linux environment on top of a variety of cache configurations. In the scope of this work, we manually assign applications to cores to take advantage of AMP configurations. Our results show that in a AMP system, different applications can benefit from various configurations to complete their work faster using less resources.

Document type: 
Thesis
File(s): 
Senior supervisor: 
Lesley Shannon
Alexandra Fedorova
Department: 
Applied Sciences:
Thesis type: 
(Thesis) M.A.Sc.

Real-time retinal blood flow imaging with Fourier Domain Optical Coherence Tomography

Author: 
Date created: 
2014-12-01
Abstract: 

Optical Coherence Tomography (OCT) is a non-invasive micrometer-resolution depth resolved medical imaging tool for diagnostic visualization of the retinal structures in vivo. In this thesis, we first describe the capability of OCT for providing structural information by quantitative retinal thickness measurements in animal models of diseases causing vision loss. In the rest of the work described in this thesis, we focus on the development of an extension of OCT called speckle variance (sv) OCT, which provides functional information such as detecting capillaries within in the retina. Retinal capillary networks are critically linked to retinal neuronal health and disease. The ability to perform accurate in vivo examination of human retinal capillary networks is therefore valuable for studying mechanisms that govern retinal homeostasis and retinal vascular diseases. A real-time implementation of the svOCT provided by the GPU acceleration was described to provide visualization of en-face vasculature networks during acquisition. A qualitative comparison study was described in this thesis by comparing the retinal vasculature images acquired from svOCT and Florescence Angiography (FA) for both healthy and diseased patients. The capability of svOCT with respect to quantifying capillary network information has been also validated in this thesis. The results of these studies suggest that this GPU accelerated svOCT has the potential to non-invasively provide useful quantitative information about human retinal capillary networks, and may have clinical and research applications for the management of retinal microvascular diseases, a major cause of vision loss worldwide. To further adapt the svOCT to be more clinical friendly, preliminary work on enhancing the real-time visualization of vascular information from distinct retinal capillary beds during acquisition was proposed and implemented in this thesis.

Document type: 
Thesis
File(s): 
Senior supervisor: 
Marinko V. Sarunic
Department: 
Applied Sciences:
Thesis type: 
(Thesis) Ph.D.

Improved Bio-inspired Artificial Gecko Adhesive by Using Hierarchical Fibrillar Structures

Author: 
Date created: 
2014-12-03
Abstract: 

Geckos are well known for being rapid climbers that have long existed in nature. The reversible and reusable adhesive on their feet intrigues scientists to explore a bio-mimetic adhesive, which inherits the adhesion properties of the gecko’s adhesives. Recent advances in electron microscopy reveal the secret of gecko’s climbing ability: there are hierarchical fibrillar structures branching from the skin of their climbing feet. Sizes of these hierarchical fibrils range from micrometer to nanometer. These fibrils are arranged to closely resemble a tree, and these tree like structures form a fibril forest on the skin of the climbing feet. Nano-fibrils in close proximity with the contacting surfaces interact with the substrate through intermolecular forces. Slender micro-fibrils extend the nano-fibrils, which are located at their open ends, to reach recesses of the contacting surfaces. The special arrangement of the fibrillar arrays enables quick attachment and detachment of the feet from surfaces of different materials and varying roughness. Inspired by the gecko’s adhesive, artificial fibrillar adhesives have been sought developing for more than a decade. Early attempts were focused on making use of the intermolecular interaction by nano-fibrillar arrays. These artificial fibrillar adhesives have achieved great performance on flat surfaces but not as good when they were used on relatively rough surfaces. Recent attempts of preparing a hierarchical fibrillar structure, which contains fibrils in different length scales, have rare success on improving adhesion performance. Evidence of extra compliancy provided by the hierarchical structure is also not clear. This thesis provides evidence that there is a correlation between structure compliancy and adhesion performance of a hierarchical fibrillar adhesive. Improved compliancy and adhesion forces are observed on a hierarchical fibrillar structure with achievements of several milestones, which include developing methods for preparing and characterizing hierarchical fibrillar structures. Experimental results also reveal the interaction of fibrillar arrays with the contacting surfaces. Information obtained is valuable for future development and application of such artificial fibrillar adhesive.

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

Method of measuring nitric oxide release by vascular endothelial cells grown in microfluidic channels

Date created: 
2014-11-21
Abstract: 

In this thesis, a simple and versatile method is developed which enables detection of nitric oxide (NO) released from vascular endothelial cells (ECs) cultured in microfluidic structures. The new culturing system and NO measurement method allow cell shape to be controlled in a non-invasive manner using microfluidic structures while NO release is monitored for cell shape versus function studies. The culturing system consists of arrays of polydimethylsiloxane (PDMS) fluidic channels 120 micrometers in depth and ranging from 100 micrometers to 3 mm in width. The number of channels in each array is varied to yield a constant cell culture surface area (75 mm2) independent of channel width. The channel surfaces are collagen-coated and ECs are cultured to confluence within the channels. A cell scraper is then used to scrape extraneous cells cultured between channels, and NO measurements are made 18 to 24 hours later. A chemiluminescence-based sensor system (NOA 280i, Sievers NO Analyzer) is utilized to measure sample NO. Initial results indicate that NO concentrations can be measured from different microfluidic channel-containing samples using this method. Initial results suggest that there is no significant difference in NO concentration derived from channels of different widths even though the degree of cell elongation varies due to physical constraint by microfluidic channel walls. Cells treated with TNFα appear more elongated and release slightly more NO than untreated cells in fluidic channels; however, the ANOVA test indicates that this difference may not be significant for both microfluidic channels or fluidic channels and forms a basis for further study.

Document type: 
Thesis
File(s): 
Senior supervisor: 
Bonnie L. Gray
Department: 
Applied Sciences:
Thesis type: 
(Thesis) M.A.Sc.

Beamforming for multiuser MIMO systems

Date created: 
2014-09-05
Abstract: 

Wireless communications systems use a multiple user scheme such as time- or frequency-division,but these do not allow truly simultaneous use of the spectrum. By deploying multiple antennas andbeamforming, it is possible, in principle, for users to share the spectrum simultaneously, and thisscenario is called the multiuser MIMO interference channel. This thesis presents new beamfomingdesign methods for this channel, derived from the convergence criteria for multi-objective optimization.Beamforming is proven to be possible for any combination of communications objectivefunctions such as mean-square error, signal-to-interference plus noise ratio, and leakage interference.Relationships are found between the number of users and number of antennas, for different objectivefunctions. The existence of a Nash equilibrium is guaranteed and the important networkingproperties of quality of service and fairness among users are accounted for. A new optimizationalgorithm, which is an extension of alternating optimization, is formulated for the design process.Its advantage over existing approaches is its significantly lower computational complexity. Severaloptimized, multi-user OFDM systems are formulated and demonstrated by simulation using statisticalchannel models in a multipath environment. The feedback overhead required for deploying thebeamforming is quantified, showing the trade-off among complexity, minimum number of antennasrequired, error performance, capacity, feedback rate, and the ability to extract multi-path diversityfor multiple users. When one of the users has priority access to the spectrum, the channel takes ona form of cognitive radio. This scenario is formulated as an optimization which requires solution viaan evolutionary algorithm, and convergence is shown to be faster when more antennas are deployed.Finally, an architecture is presented that enables a secondary (i.e., low priority) user, whose terminalscannot directly ”see” each other, to communicate in the presence of multiple primary users.The cost is the need for all the primary users to be modified to collaborate with the secondary user,and for several MIMO relays to be installed. The secondary capacity is maximized under constraintsof transmission power and interference to the primary receivers, and relay selection. This conceptshowcases several communications techniques including eigen-beamforming, channel selection andcapacity optimization.

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

Architecture and Design of a Low Frequency Mechanical Actuation Device for Pre-Hospitalization Treatment of Myocardial Infarction Patients

Date created: 
2013-06-20
Abstract: 

One of the major factors in increasing the survival rate of patients suffering from acute coronary ischemia is the speed of intervention. Two major techniques are currently in use: pharmacological and interventional. The former is slow acting and often leads to incomplete reperfusion, while the latter requires specialized personnel and a cathlab. This thesis proposes a novel method for pre-hospitalization treatment of patients with acute coronary ischemia that can be safely applied by a minimally trained individual prior to or during patient transportation to hospital. It consists of applying low frequency mechanical vibration to the left intercostal space of patient’s chest during diastole to induce vibration on the heart and thus on the coronary arteries. Additionally, the method includes application of direct, distal, mechanical, arterial deformation to induce turbulence in the blood flow. Mechanical vibration increases coronary blood flow, acts as a strong vasodilator, and relieves heart spasms likely to be seen in heart attack patients. Direct arterial deformation generates turbulence in the blood, which amplifies mixing of clot busting agents with thrombi. Furthermore, it imposes shear stress on the clot wall to achieve clot displacement and/or disruption. In order to investigate the impact of mechanical actuation on clot lysis and also to examine feasibility of a device for application of mechanical vibration on the chest of myocardial infarction patients, three major studies are presented that also form the three objectives of this work: the first study introduces an electromechanical apparatus to study the effects of mechanical vibration and deformation on disrupting blood clots in-vitro with and without combined use of thrombolytic agents. The second study describes the design and architecture of a prototype device for application of Diastolic Timed Vibration (DTV) on the chest of heart attack patients for increasing the coronary blood flow, and improving the weak relaxation of the ill myocardium. The final experiment presents a preliminary investigation on a human subject to determine whether direct distal arterial deformation could cause turbulence in the blood flow for better mixing of fibrinolytics with clot and induce stress on the clot and the clot wall.

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

Automatic Methods for Human Embryo Component Extraction

Author: 
Date created: 
2014-08-18
Abstract: 

In this thesis, we proposed two novel methods for blastomere extraction and trophectoderm segmentation in an attempt to aid physicians in determining embryo’s viability. Accurate assessment of embryo’s viability can play a vital role towards optimization of in-vitro fertilisation (IVF) treatment outcomes. The first proposed automatic method is developed to identify blastomeres in human embryo HMC (Hoffman Modulation Contrast) images of day-1 to day-2. Our algorithm applies isoperimetric graph partitioning, followed by a novel region merging algorithm to approximate blastomeres positions. Ellipsoidal models are then used to approximate the shape and the size of each blastomere. The proposed algorithm is evaluated on a dataset of 40 embryo images and it exhibits an average blastomere extraction accuracy of 80%. The second method segments Trophectoderm (TE) regions in embryos of day-5 (also known as blastocysts) by first eliminating the inhomogeneities of the blastocysts surface using Retinex theory. A level set algorithm is then used to segment TE regions. We have tested our method on a dataset of 85 images and have achieved a segmentation accuracy of 85% for grade A, 89% for grade B and 92% for grade C.

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

Visual Tracking and Region Alignment in Surgical and Medical Education

Author: 
Date created: 
2014-08-15
Abstract: 

To design a user computer interactive system for medical and surgical education, a study of visual tracking and region alignment is presented in this thesis. To effectively track the surgical instrument in the view of an endoscope, both the Gaussian type tracking methods, based on the Extended Kalman Filter (EKF) and the Adaptive Gaussian Mixture Model (AGMM) and the non-Gaussian type Particle Filter (PF) approach are proposed and evaluated using the video data captured from both in-vitro and in-vivo environments. To improve the stability of instrument tracking, a hybrid method integrating PF and AGMM is presented. One of the extensions of the visual tracking results is to match the same region under different viewing conditions and combine it with Augmented Reality (AR). A preliminary study of user-defined region matching algorithms, based on feature descriptors such as SIFT, SURF and ORB, are proposed and integrated with AR development.

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

Haptic Data Processing for Teleoperation Systems: Prediction, Compression and Error Correction

Author: 
Date created: 
2013-06-10
Abstract: 

This thesis explores haptic data processing methods for teleoperation systems, including prediction, compression, and error correction. In the proposed haptic data prediction method, unreliable network conditions, such as time-varying delay and packet loss, are detected by a transport layer protocol. Given the information from the transport layer, a Bayesian approach is introduced to predict position and force data in haptic teleoperation systems. Stability of the proposed method within stochastic formalism is presented based on the notion of passivity-based control. In the proposed haptic data compression method, compression techniques based on fixed rate down-sampling are presented for efficient transmission over the network. Objective and psychophysical evaluations are conducted to demonstrate the compression performance of the proposed method and the adaptive down-sampling method based on human perception. By presenting the two evaluation measures, the usefulness of the objective evaluation measure for haptic data is investigated. Finally a forward error correction method is applied to haptic data over the unreliable network. Given the psychophysical evaluation results to determine the required number of bits for haptic data quantization, the error correction performance is presented under the additive noise and packet loss behavior.

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

Identification of Pacemaking Region in Zebrafish Heart from Optical Mapping Data

Author: 
Date created: 
2013-04-24
Abstract: 

Junctional Ectopic Tachycardia (JET) is a cardiac arrhythmia which occurs immediately after open heart surgery in young children. In specific populations, JET has extremely high incidence, of up to 50%. There has not been a specific mechanism elucidated by clinical data or basic science. As a widely used vertebrate cardiovascular biological model, zebrafish heart is being studied to reveal the leading reasons of JET. Optical mapping (OM) techniques provide an effective approach to observe cardiac functionality by recording zebrafish heart action potential propagation. However, processing of vast amount of OM data also poses challenges on fast and accurate processing, measurement and interpretation. This thesis presents novel automated pipelines for processing zebrafish heart OM data and identifying pacemaking regions from it through signal analysis. We first introduce a preprocessing pipeline for enhancing very low signal-to-noise ratio original OM data, which involves spatial-temporal smoothing, cycle averaging, drifting correction and scaling. After that, we present a computer assisted OM signal manually labeling pipeline, which reduces the manual workload significantly by clustering spatially adjacent similar signals. Furthermore, we make physiologically relevant measurements on OM data and do statistical analysis comparing different labeled regions. Finally, we propose a two-step signal clustering based method to divide atrium into different functional regions followed by pacemaking region identification. We present the formulation of these methods and discuss their validity and performance in various aspects. The work presented in this thesis could lead to significantly faster and larger scaled experimentation in optical mapping related physiology research.

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