About
Research Interest
- Computational Chemical Biology
- Protein-Ligand Interactions
- Medical Image Analysis
- Computational Cardiology
- Cardiovascular Biology
- Environmental Emission Modelling
This is my dog Mengmeng
(aka my research mate)
Bet u'll luv her :)
Education / Employment
| 2024 – Ongoing | Department of Biological Sciences University of Toronto |
| 2024 – Ongoing | Department of Physical and Environmental Sciences University of Toronto |
| 2022 – Ongoing | Department of Computer and Mathematical Sciences University of Toronto |
Academic Service
| 2026 | Peer-Reviewer Current Medical Imaging |
Research
Do Cat-egories Really Exist?
An animal behaviour study of domestic-cat personality structure, integrating Gaussian mixture modelling, continuous-reference simulation, resampling stability, cross-algorithm comparison, and demographic sensitivity to test whether reproducible profiles represent discrete behavioural categories.
Release Sensitivity in Biomedical Knowledgebases
A support-aware, reproducible audit of release sensitivity across Open Targets Platform versions, assessing changes in disease–target scores, fixed-roster rankings, candidate turnover, and entity mapping.
Dynamic FABP Recognition of PFECAs
A computational chemical biology study of PFECA-FABP molecular recognition, integrating chemical, structural, and protein-family evidence to investigate cross-isoform binding behaviour.
Multimodal Lac Operon Profiling
A multi-assay characterization of E. coli lac operon regulation, integrating β-galactosidase activity, chromogenic phenotyping, protein profiling, and colony PCR to distinguish wild-type and mutant regulatory states.
Structured Coronary Representation Framework
A structured, robustness-aware framework for interpretable CCTA-derived coronary representation and multi-source (CCTA, DSA, collaboration with hospitals) clinical validation.
Release Sensitivity of Historical EDGAR Emissions
A lineage-aware, reproducible audit of historical CH4 and N2O revisions across EDGAR releases, assessing release sensitivity in emission magnitudes, long-term trends, and country-sector rankings.
Controlled Organometallic Catalysis
An organometallic chemistry study of transition-metal catalysis, integrating Pd/Cu redox cycling, Co–C bond activation, radical mechanisms, and structural evidence to examine how catalytic environments control reactivity and selectivity.
Leakage-Controlled Emission Modelling
An environmental informatics study of European Cd, Hg, and Pb emission inventories, integrating data harmonization, leakage-controlled machine learning, spatial validation, and model benchmarking to evaluate short-horizon inventory updating.
Comprehensive Chemical Assessment of Rhus typhina Silage for Herbivores at the Toronto Zoo
A QA / QC-focused proposal integrating inorganic nutritional quality and fermentation markers.
Quinine Fluorescence and Quenching
A fluorescence spectroscopy study of quinine, integrating quantitative calibration, chloride quenching, Stern-Volmer analysis, and pH-dependent fluorescence to characterize emission behaviour across changing chemical environments.
Nucleotide Profiling in Infant Formula
An analytical chemistry study of nucleotide composition in infant formula, integrating UV–Vis calibration, HPLC separation, matrix-aware sample preparation, and chromatographic quantitation of five nucleotide monophosphates.
Systolic Blood Pressure Risk Factor Modelling
A biostatistics study of systolic blood pressure variability, integrating multivariable regression, nested model comparison, interaction analysis, and diagnostic testing to characterize demographic, lifestyle, and treatment-related associations with SBP.
Occlusion-Robust Object Recognition
A computer vision study of overlapped-object recognition, integrating local SIFT features, scale-space keypoint analysis, and part-level visual representations to improve weakly supervised recognition under occlusion.
Nonlocal Mean Fracture Imaging
An image processing study of low-contrast CT fracture recognition, integrating background normalization, non-local mean filtering, threshold-based region separation, and graph-based denoising to improve crack extraction under high-noise conditions.
Publication
A network update redistributes functional content and candidate priorities
A reproducible, support-aware technique for evaluating how network knowledgebase updates alter functional capture, candidate-budget allocation and donor-specific priorities while distinguishing aggregate performance shifts from functional composition, structural baselines and identity-linked uncertainty.
Behavioural profiles in domestic cats: repeatable trait combinations and method-dependent membership
A reproducible, evidence-linked analysis of personality profiles in domestic cats while distinguishing profile-level separation and repeatability from analytical sensitivity and individual membership uncertainty.
TRACE: Evidence-Linked Computational Representation of Coronary Quantification from CCTA to Clinical Reports
A reproducible, evidence-linked framework for quantifying coronary anatomy and stenosis from CCTA while preserving measurement provenance, assessability, perturbation sensitivity, and cross-source correspondence with clinical reports.
Support-aware evaluation of release sensitivity in a biomedical knowledgebase
A reproducible, support-aware framework for quantifying how Open Targets Platform release changes affect disease–target scores, rankings, candidate turnover, and downstream analytical conclusions.
Historical greenhouse gas trends and country rankings can depend on database release
A reproducible release-sensitivity framework for quantifying how historical EDGAR revisions affect greenhouse-gas trends, country rankings, and analytical conclusions.
CoroConcept: Three-Dimensional Image Processing for Auditable Coronary Descriptor Extraction from CCTA Segmentation Masks
An auditable 3D post-segmentation pipeline that turns CCTA coronary masks into structured quantitative descriptors.
Leakage-Controlled Machine Learning for European Cd, Hg, and Pb Emission Inventories
An auditable, leakage-controlled machine-learning framework for provisional spatial updating of European Cd, Hg, and Pb emission inventories.
Recognition of Overlapped Objects via Weakly-Supervised Methods with Bag of Alphabets
A weakly-supervised recognition framework using a bag-of-alphabets model to identify and classify overlapping objects.
Nonlocal Mean Filtering Algorithm for Low Contrast Images and its Application
An improved nonlocal mean filtering approach for enhancing low-contrast image quality and demonstrating its practical applications.
Contact
Email: