Sydney eScholarship
Not a member yet
31878 research outputs found
Sort by
Russian “Otherness” - Barriers and Solutions to Performing Russian Opera and Art Song for English-Speaking Classical Singers
No full textRussian classical vocal music is acknowledged as part of the canon, yet Russian song repertoire is rarely performed in English-speaking Conservatories. This mixed-method research identifies the barriers English-speaking classical singers experience in learning Russian vocal music. The study uses semi-structured interviews, additional sources analysis, and autoethnography to identify the causes and propose solutions to this problem.
Based on interviews with professional operatic and lieder singers, classical singing students and voice teachers, we identified three key pillars of barriers. Unexpectedly, the language barrier and unfamiliar Cyrillic alphabet, while being significant obstacles, are not the most critical barriers singers have to overcome in accessing and performing Russian classical song.
The singular most surprising barrier found in this research is the overall lack of familiarity with Russian culture and, for some participants, a perception of this culture as ‘alien’. Second to this is the lack of familiarity with Russian classical vocal music among singers, coaches, and, in particular, voice teachers who play a crucial role in repertoire selection for young singers.
There was previously a lack of comprehensive resources available for singers needing training in Russian lyric diction. This research found a lack of awareness of the existing resources among singers and voice teachers.
The creative work, a lecture-recital on Russian coloratura repertoire supports our findings by including Early Romantic Russian music, styles of performance, music suitable for lighter voices, younger singers, and coloratura voices, gaps in thinking about Russian music are explored and demonstrated through preparation and performance of a Russian song recital.
This research addresses the barriers to engagement with Russian repertoire, and advises Conservatories on ‘normalizing’ Russian repertoire for voice students by creating a supportive environment for them to explore it
Revealing the evolution of viruses using phylogenetics, metagenomics and structural prediction
No full textMetagenomic sequencing, particularly total RNA sequencing (i.e., meta-transcriptomics), has emerged as a powerful tool for virus discovery. Although the known virosphere has expanded dramatically in recent years, the viruses associated with much of the tree of life remain largely undiscovered, limiting our ability to infer the evolutionary and ecological processes that shape viral diversity. The central aim of my thesis was to explore viral diversity in under-sampled species critical to answering fundamental questions about virus evolution and ecology. To achieve this, I combined metatranscriptomic sequence data with state-of-the-art phylogenetic and structural analysis. I determined a novel papillomavirus, a likely aetiological agent of oral papilloma-like lesions in a fur seal, and characterised the virome of Australian dogs, including the first RNA virus in dingoes. I also conducted broad virus discovery surveys of publicly available transcriptome data sets. By analysing species from across the breadth of the plant kingdom I identified 104 potentially novel viruses, including sequences divergent enough to represent novel virus families, as well as the first virus genomes in bryophytes and lycophytes. Through an analysis of flaviviruses across the Metazoa I discovered 32 novel flavivirus sequences. These included an orthoflavi-like virus in a coral, marking the earliest diverging metazoan host, and a substantial expansion of the host range of pestiviruses. I next used the expanded Flaviviridae sequence data set to investigate glycoprotein distribution within the family. Through structural homology searches I uncovered a complex evolutionary history for the Flaviviridae, characterised by bacterial gene capture and inter-genus recombination. In sum, the research in this thesis expands the known virosphere and highlights how novel and highly divergent viruses can be key to uncovering the fundamental mechanisms, patterns, and processes of virus evolution
Medical Image Segmentation under Data-Efficient Scenarios
No full textMedical image analysis is essential for accurate diagnosis, treatment planning, and patient management. Among its tasks, segmentation is key to identifying anatomical structures and abnormalities. Although deep learning has significantly advanced segmentation by learning complex patterns from large datasets, it requires vast amounts of labeled data, which is expensive and time-consuming to obtain. Consequently, transfer learning has emerged to bridge distribution gaps and reduce reliance on extensive labeled data. Nonetheless, challenges remain, such as semantic gaps in image categories, limitations of adversarial learning, and issues with generalizing to unseen target data. This thesis presents several deep learning-based domain adaptation methods for data-efficient medical image segmentation:
The first part of this thesis introduces CAPL-Net, a novel deep unsupervised domain adaptation (UDA) approach specifically designed for cross-domain nuclei instance segmentation and classification using category-aware feature alignment and pseudo-labeling.
The second contribution presents an advanced dynamic diffusion-based framework (DDF-UDA) for cross-domain optic disc (OD) and optic cup (OC) segmentation in fundus images. DDF-UDA includes an adaptive module based on feature- and pixel-level diffusion processes, along with a Nash equilibrium-based adjustment strategy to reduce cross-domain discrepancies.
Finally, we propose a novel zero-shot domain adaptation (ZSDA) paradigm, named LMaD, to address the challenge of requiring access to target images during training. LMaD utilizes a language model as a feature disentangler for ZSDA medical image segmentation tasks, aiming to eliminate interference caused by domain-specific information, resulting in superior performance.
Overall, this thesis advances UDA and ZSDA methods in medical image segmentation, offering practical solutions to overcome data scarcity and enhance model generalization in diverse clinical settings
Bridging the Commercialization Gap for the Direct Air Capture of Atmospheric CO2
No full textThe present work is a systematic investigation that addresses the primary concerns associated
with scaling Direct Air Capture (DAC) technology. Chapter 2 provides a non-technical
communication about the value of DAC relative to other forms of Carbon Dioxide Removal
(CDR). If CDR systems are compared on a simple cost basis, DAC will be more expensive per
tonne of CO2 due to technology’s nascency relative to nature-based forms of CDR such as
afforestation/reforestation. Chapter 2 reasons that a simple accounting framework neglects the
inherent value of long-duration CO2 sequestration, which is a possibility when DAC is paired
with carbon mineralization or geological sequestration but is not possible with
afforestation/reforestation. Chapter 2, therefore, argues for the implementation of a more
complex pricing framework that accounts for the value of long duration storage in CDR
solutions.
Chapter 3 begins this thesis’ examination of chemical sorbents for DAC. The chapter describes
a novel, green mechanochemical synthesis for a zirconium-based metal-organic framework
(MOF), UiO-66-NH2, which was believed to be a potential sorbent for DAC. UiO-66-NH2 was
synthesized at a rate of 4.52 g per 90 minutes with a 43% yield at a levelized cost of $6,498/kg
of MOF. It became clear through CO2 gas sorption results, however, that the MOF itself was
not a sufficient adsorbent material. As a result, Chapter 4 describes the mechanochemical
impregnation of UiO-66-NH2 with a redox active guest, 9,10-phenanthrenequinone. The
composite material exhibited the desired electrochemical behavior after a rapid 30-minute
mechanochemical impregnation where the redox-active guest comprised approximately 13%
of the composite material’s mass. This chapter demonstrates that mechanochemical
impregnation is a rapid functionalization technique that can be leveraged with other porous
materials for many industrial applications including and beyond CO2 capture. Overall, Chapter
4 examines this novel functional material as a solid-state cathode for electrochemical CO2
capture, and the chapter demonstrates that mechanochemical functionalization is a useful
technique for rapidly introducing desired chemical properties for various industrial
applications.
After examining solid-state materials for DAC in Chapters 3 and 4, Chapter 5 is a technological
pivot away from solid-state sorbents toward a solution-state electrochemical DAC process.
iv
Chapter 5 describes the investigation of several types of archetypal redox active organic
molecules—such as diazines and quinones—for a continuous, non-aqueous electrochemical
CO2 capture process. The highest performing organic molecule was phenazine, which
maintained a 100% coulombic efficiency over 9.5 hours of testing with a theoretical minimum
energy of 77.2 kJ/mol of CO2 captured. Additionally, this chapter describes the development
of a 3D printed zero-gap electrochemical flow cell, which was designed to reduce the
engineering costs associated with facilitating a continuous electrochemical reaction. In Chapter
6, the technology further evolves toward a green, aqueous electrochemical DAC process. The
system demonstrated reversible electrochemical behavior over 100 cycles or 205 hours and
maintained an average coulombic efficiency of 100% and an average capacity retention of
99.8%. Additionally, the system has an estimated theoretical minimum energy of 24.6 kJ/mol.
Chapter 6 also includes further optimization of the electrochemical flow cell, which was
performed with computational fluid dynamics (CFD) software to improve the flow of solution
to the surface of the electrode, and it also includes a techno-economic analysis (TEA) to analyze
variables that facilitate the commercialization of a redox flow DAC process
An investigation of the environmental health risks from rural sanitation infrastructure: with a case study from Fiji
No full textAccess to safely managed sanitation has emerged as a global challenge. Nearly half the global population lives in rural areas and generally relies on on-site sanitation systems. Sustainable Development Goal target 6.2 aims to achieve adequate and equitable sanitation by 2030. Currently, all regions are off-track in achieving the target due to the inequalities between the regions and rural and urban areas. Assessing sanitation back-end containment safety is a continuing challenge due to the unquantified variability of types of sanitation back-end infrastructure present in rural communities. This research aimed to investigate the environmental health risks from on-site sanitation back-ends in rural communities with a case study in Fiji (Western Pacific region). The Fijian case study involved 29 rural communities from five water catchments across the three main islands. The results from the sanitation observation data showed that 58% of the studied households used uncontained plastic or metal tanks as back-ends. However, 51 – 64% of the studied households inaccurately reported uncontained tank-type back-ends as septic systems. The over-reporting of septic systems highlighted the underlying challenges with data accuracy and the need to provide clear categorisation of sanitation back-ends. Lateral transport of the contaminants was limited to 10 m from the sampled back-ends, predominantly due to the silt and clay-type soils. This significant finding highlighted the protective nature of soil with lower permeability in restricting the lateral transport of faecal sludge contaminants. A novel distance and soil-type-based sanitation risk assessment framework was developed. The framework was then applied to a case study in Timor-Leste (South-East Asian region) to assess its applicability. This research significantly contributed to the understanding of back-end containment safety. These findings will contribute to advancing the progress towards achieving SDG target 6.2
Chasing P2X7 Receptor Ligands for Cardiovascular Disease
No full textCardiovascular disease (CVD) remains the leading cause of death in Australia, with atherosclerosis
driving coronary artery disease (CAD), a severe form of CVD. The persistent inflammatory
component of CAD necessitates novel therapies, as current treatments like low-dose colchicine rely
on broad immunosuppression with inherent risks. P2X7R antagonists offer a promising alternative by
selectively targeting CAD-specific inflammatory pathways, benefiting from extensive prior clinical
evaluation and a favourable safety profile.
This Thesis aimed to expand the structure-activity relationship (SAR) understanding of P2X7R
antagonists and advance their preclinical development. Three compound series were investigated:
PKT100 (trifluoroadamantyl benzamide), PKT200s (heterocyclic adamantyl cyanoguanidines),
adamantyl–quinolinyl linkers, and the novel PKT400s (5-alkoxyalkylamine trifluoroadamantyl
benzamides). The PKT200 chemotype was extensively explored, with ortho-halogenation emerging
as a strategy to enhance potency. Adamantyl–quinolinyl pendant groups on various cores led to the
discovery of a potent malononitrile analogue and insights into sulfonylguanidine and pyridazin-3-one
scaffolds.
A systematic preclinical assessment evaluated PKT100s, PKT200s, and PKT400s. While PKT200s
exhibited exceptional potency, poor metabolic stability limited clinical potential. However, PKT400s
demonstrated promising in vitro efficacy and pharmacokinetic (PK) profiles, outperforming PKT100
and the benchmark AZD9056.
These findings contribute to the development of P2X7R antagonists with improved pharmacokinetic
and pharmacodynamic properties, advancing therapies for CAD-associated inflammation
MaaS and Sustainability: How MaaS may deliver sustainable goals in an urban context
No full textMobility-as-a-Service (MaaS) has gained attention as an innovative approach to enhancing mobility
services and promoting sustainable travel by reducing private car dependency. Despite its potential,
MaaS faces challenges in encouraging sustainable travel behaviour and achieving commercial
scalability. This thesis addresses these challenges through two studies focusing on environmental
sustainability and commercial viability. The first study analyses MaaS users’ mode choice behaviour
using revealed preference data from the Sydney MaaS trial. A joint choice model was developed to
examine how different MaaS products influence sustainable travel behaviour change in terms of
mode selection. Findings suggest that well-designed MaaS bundles, particularly those with financial
incentives and multiple mobility options, can reduce private vehicle use and encourage shared and
public transport. However, the study also highlights the need for financial incentives and the potential
for unintended travel behaviour changes.
The second study explores the commercial potential of expanding MaaS into a multiservice platform
integrating non-mobility services to achieve scalability. Semi-structured interviews with stakeholders
identified services for the Australian market, including entertainment discounts, delivery services,
media subscriptions, and point rewards schemes. A stated choice experiment was conducted to
evaluate user preferences for multiservice offers. Discrete choice modelling revealed that pay-asyou-
go multiservice options are preferred, with demand for non-mobility services varying by market
segment. Willingness-to-pay estimates provide insights into the feasibility of integrating these
services into MaaS. This thesis offers policy implications for MaaS design and contributes with
original empirical evidence on sustainable travel behaviour under MaaS and public preferences for
multiservices, providing insights into achieving commercial scalability
Communicating health research to a diverse audience: what can plain language summaries offer?
No full textPlain language summaries (PLSs) are research summaries written with a non-expert audience in mind. PLSs have been a part of academic publishing for several decades and serve as a means of providing accessible research to a diverse audience. Mainly located in journal articles in text-only format, PLSs also form part of research funding applications for consumer groups or government agencies and can be found in audio, visual and graphical formats. Although growing in popularity in recent years with journal publishers and researchers, most health and medical journals still do not publish PLSs. Of those that do, they are not a mandatory requirement. PLSs play a valuable role in communicating health research to a diverse audience, however they remain an underutilised resource.
This thesis demonstrates where gaps exist in the collective understanding of how PLSs could be optimised to communicate health research to the public. Such gaps exist in evaluating and challenging assumptions related to current practices in producing, publishing and disseminating PLSs. Most of the work in the field of PLSs has been experimental in design, with less attention given to the adoption, implementation and sustainability of PLSs. Focus on these areas is vital as PLS application into practice remains inconsistent. PLS have the power to meaningfully engage people with trustworthy research, playing an essential role in the oftentimes confusing landscape of health information. Attention must be given by scholarly publishing and academia to support researchers to write and distribute PLSs, providing access to a diverse audience of people
Habitability dynamics of Pacific Islands in a changing climate
No full textIslands and atolls in the Pacific face heightened climate risk due to low elevations and limited resources. The question of (unin)habitability in these locations is often simplified to characteristics of hazard exposure, reinforcing assumptions of inevitable mass migration. This thesis builds on existing research, challenging the idea of linear causality between hazard exposure and migration by examining the relative and dynamic nature of habitability in the Pacific.
The compounding and cascading effects of climate change add complexity in untangling how impacts influence habitability perceptions and migration decisions. I began this work by systematically identifying the multicausal factors that affect climate-affected populations’ decisions to leave or remain in place. I then engaged experts to contextualise these findings by hazard type and to the specific geography of Pacific atolls, using the country of Kiribati as a case study location. Participatory model building activities were used to understand habitability needs and migration decision-making in Kiribati, resulting in a systems map of habitability.
This thesis emphasises co-produced knowledge with research partners, answering calls for a grounded understanding of habitability from perspective of those with localised knowledge and experience. The work in this thesis culminates in a locally derived system dynamics model of the social-ecological system in Kiribati over the next century. The novelty of this work lies in the identification of key drivers of habitability in Kiribati and examination of how these core feedback loops shift in dominance in the coming decades. The results of this work have practical implications for the strategic adaptation planning in the Pacific and can be used to further informed, evidence-based policy development
Proximal sensing surveys (EM & Gamma) across 36 farms from the grain growing regions of Australia
No full textProximal soil electromagnetic induction (EM) & gamma radiometric (gamma) surveys from 36 farms across Australia's dryland cropping regions. Surveyed fields correspond to those sampled in the '3D PAWC and constraint mapping' project. Soil proximal surveys were compiled between 2023 and 2025.
These surveys were collected to be used as modelling covariates in digital 3D soil maps that describe on-farm soil constraints and PAWC in a given area and can be used to inform management decisions. These have been collected at many sites across Australia to give an idea of the investment value for proximal surveys in different environments and soil types.
Surveyed areas range from 200 - 2,000 ha at specific farms, while the total surveyed area represented in this dataset exceeds 26,000 ha of dryland cropping soils.
Proximal surveys were collected at 36 m swathe widths using gator mounted EM & gamma sensors.
EM & gamma surveys are stored seperately in .csv format for each farm.
EM data consists of 4 x depth slices to a maximum of 300 cm from a DualEM sensor.
Gamma data collected by an RSX-1 gamma-ray spectrometer consists of Potassium, Uranium & Thorium concentrations as well as total count.
Coordinates are present in 'EPSG:4326' projection.
Farms were surveyed between 2023 and 2025. The date of survey is available for all farms.
GRDC Project UOS2206-009RTX
The datasets with raw values are stored on the USYD-RDS at \\shared.sydney.edu.au\research-data\PRJ-MLCons. This data has restricted access and the data is de-identified as it contains information from private properties. Third-parties may be able to access the data subject to terms agreed to by the GRDC and the University of Sydney and/or a data supply and licence agreement. Please contact Dr Patrick Filippi ([email protected]) to request access to the data