arXiv subjects

Kelvin Lee

Publications and source records attributed to Kelvin Lee.

7 recordsLinked to original sources

arXiv2026

Same Branches, Different Trees: A Bifurcation Connectedness Metric for Coronary Artery Segmentation and FFR-CT Decision Agreement

Fractional flow reserve derived from CT angiography (FFR-CT) simulates flow through a patient-specific vessel model, so its accuracy depends on the connectedness of the segmented tree, not only on volumetric overlap: a segmentation can reach high Dice yet sever a bifurcation, dropping the downstream subtree and reversing the treatment decision. Topology-aware losses such as clDice and Skeleton Recall act on the global centreline and can miss localised breaks. We study the Bifurcation Connectedness Score (BCS), which scores connectedness at each ground-truth bifurcation, and soft-BCS, its differentiable training surrogate. BCS captures a property of segmentation quality the standard metrics miss: it responds strongly to breaks in connectedness while staying largely unchanged under connectedness-preserving narrowing. Higher BCS accompanies closer agreement between the FFR-CT decisions a solver makes on predicted versus ground-truth geometry, most clearly in severe disease (OR 2.16, CI [1.23, 4.18]). Both decisions come from the same solver, so this reflects geometric, not clinical, fidelity. In training, soft-BCS and Skeleton Recall recover the same branches but build different trees. Recovering branches and keeping them connected are separable properties, so we recommend reporting a measure of each.

cs.CV↗

arXiv2026

STEPIC: High-Speed Imaging via Spatio-Temporal Encoding in Photonic Integrated Circuits

High-speed imaging of cells in flow is essential for probing cellular heterogeneity in large populations. Existing imaging approaches based on single-pixel detection and spatio-temporal encoding provide exceptional speed, but typically rely on bulky free-space optics, long dispersive elements, and are prone to alignment instabilities. Here, we introduce STEPIC Microscopy, the first fully integrated on-chip system for high-speed imaging via spatio-temporal encoding in photonic integrated circuits. Our platform leverages waveguides, splitters, fiber delay-lines, and 3D optical remappers to encode spatial information into the temporal domain, enabling robust image reconstruction of cells flowing through microchannels. The monolithic architecture provides a compact and robust platform for high-throughput bioimaging, enabling scalable and practical implementations of ultrafast imaging systems.

physics.optics↗

arXiv2025

A Modular, Data-Free Pipeline for Multi-Label Intention Recognition in Transportation Agentic AI Applications

In this study, a modular, data-free pipeline for multi-label intention recognition is proposed for agentic AI applications in transportation. Unlike traditional intent recognition systems that depend on large, annotated corpora and often struggle with fine-grained, multi-label discrimination, our approach eliminates the need for costly data collection while enhancing the accuracy of multi-label intention understanding. Specifically, the overall pipeline, named DMTC, consists of three steps: 1) using prompt engineering to guide large language models (LLMs) to generate diverse synthetic queries in different transport scenarios; 2) encoding each textual query with a Sentence-T5 model to obtain compact semantic embeddings; 3) training a lightweight classifier using a novel online focal-contrastive (OFC) loss that emphasizes hard samples and maximizes inter-class separability. The applicability of the proposed pipeline is demonstrated in an agentic AI application in the maritime transportation context. Extensive experiments show that DMTC achieves a Hamming loss of 5.35% and an AUC of 95.92%, outperforming state-of-the-art multi-label classifiers and recent end-to-end SOTA LLM-based baselines. Further analysis reveals that Sentence-T5 embeddings improve subset accuracy by at least 3.29% over alternative encoders, and integrating the OFC loss yields an additional 0.98% gain compared to standard contrastive objectives. In conclusion, our system seamlessly routes user queries to task-specific modules (e.g., ETA information, traffic risk evaluation, and other typical scenarios in the transportation domain), laying the groundwork for fully autonomous, intention-aware agents without costly manual labelling.

cs.LG↗

arXiv2021

On the Quality Requirements of Demand Prediction for Dynamic Public Transport

As Public Transport (PT) becomes more dynamic and demand-responsive, it increasingly depends on predictions of transport demand. But how accurate need such predictions be for effective PT operation? We address this question through an experimental case study of PT trips in Metropolitan Copenhagen, Denmark, which we conduct independently of any specific prediction models. First, we simulate errors in demand prediction through unbiased noise distributions that vary considerably in shape. Using the noisy predictions, we then simulate and optimize demand-responsive PT fleets via a linear programming formulation and measure their performance. Our results suggest that the optimized performance is mainly affected by the skew of the noise distribution and the presence of infrequently large prediction errors. In particular, the optimized performance can improve under non-Gaussian vs. Gaussian noise. We also find that dynamic routing could reduce trip time by at least 23% vs. static routing. This reduction is estimated at 809,000 EUR/year in terms of Value of Travel Time Savings for the case study.

stat.ML↗

arXiv2021

Atomium: The astounding complexity of the near circumstellar environment of the M-type AGB star R Hydrae. I. Morpho-kinematical interpretation of CO and SiO emission

Evolved low- to intermediate-mass stars are known to shed their gaseous envelope into a large, dusty, molecule-rich circumstellar nebula which typically develops a high degree of structural complexity. Most of the large-scale, spatially correlated structures in the nebula are thought to originate from the interaction of the stellar wind with a companion. As part of the Atomium large programme, we observed the M-type asymptotic giant branch (AGB) star R Hydrae with ALMA. The morphology of the inner wind of R Hya, which has a known companion at ~3500 au, was determined from maps of CO and SiO obtained at high angular resolution. A map of the CO emission reveals a multi-layered structure consisting of a large elliptical feature at an angular scale of ~10'' that is oriented along the north-south axis. The wind morphology within the elliptical feature is dominated by two hollow bubbles. The bubbles are on opposite sides of the AGB star and lie along an axis with a position angle of ~115 deg. Both bubbles are offset from the central star, and their appearance in the SiO channel maps indicates that they might be shock waves travelling through the AGB wind. An estimate of the dynamical age of the bubbles yields an age of the order of 100 yr, which is in agreement with the previously proposed elapsed time since the star last underwent a thermal pulse. When the CO and SiO emission is examined on subarcsecond angular scales, there is evidence for an inclined, differentially rotating equatorial density enhancement, strongly suggesting the presence of a second nearby companion. The position angle of the major axis of this disc is ~70 deg in the plane of the sky. We tentatively estimate that a lower limit on the mass of the nearby companion is ~0.65 Msol on the basis of the highest measured speeds in the disc and the location of its inner rim at ~6 au from the AGB star.

astro-ph.SR↗

arXiv2020

Atomium: A high-resolution view on the highly asymmetric wind of the AGB star Pi1 Gruis. I. First detection of a new companion and its effect on the inner wind

The nebular circumstellar environments of cool evolved stars are known to harbour a rich morphological complexity of gaseous structures on different length scales. A large part of these density structures are thought to be brought about by the interaction of the stellar wind with a close companion. The S-type asymptotic giant branch star Pi1 Gruis, which has a known companion at ~440 au and is thought to harbour a second, closer-by (<10 au) companion, was observed with the Atacama Large Millimeter/submillimeter Array as part of the ATOMIUM Large programme. In this work, the brightest CO, SiO, and HCN molecular line transitions are analysed. The continuum map shows two maxima, separated by 0.04'' (6 au). The CO data unambiguously reveal that Pi1 Gru's circumstellar environment harbours an inclined, radially outflowing, equatorial density enhancement. It contains a spiral structure at an angle of 38+/-3 deg with the line-of-sight. The HCN emission in the inner wind reveals a clockwise spiral, with a dynamical crossing time of the spiral arms consistent with a companion at a distance of 0.04'' from the asymptotic giant branch star, which is in agreement with the position of the secondary continuum peak. The inner wind dynamics imply a large acceleration region, consistent with a beta-law power of ~6. The CO emission suggests that the spiral is approximately Archimedean within 5'', beyond which this trend breaks down as the succession of the spiral arms becomes less periodic. The SiO emission at scales smaller than 0.5'' exhibits signatures of gas in rotation, which is found to fit the expected behaviour of gas in the wind-companion interaction zone. An investigation of SiO maser emission reveals what could be a stream of gas accelerating from the surface of the AGB star to the companion. Using these dynamics, we have tentatively derived an upper limit on the companion mass to be ~1.1 Msol.

astro-ph.SR↗

arXiv2019

Online Predictive Optimization Framework for Stochastic Demand-Responsive Transit Services

This study develops an online predictive optimization framework for dynamically operating a transit service in an area of crowd movements. The proposed framework integrates demand prediction and supply optimization to periodically redesign the service routes based on recently observed demand. To predict demand for the service, we use Quantile Regression to estimate the marginal distribution of movement counts between each pair of serviced locations. The framework then combines these marginals into a joint demand distribution by constructing a Gaussian copula, which captures the structure of correlation between the marginals. For supply optimization, we devise a linear programming model, which simultaneously determines the route structure and the service frequency according to the predicted demand. Importantly, our framework both preserves the uncertainty structure of future demand and leverages this for robust route optimization, while keeping both components decoupled. We evaluate our framework using a real-world case study of autonomous mobility in a university campus in Denmark. The results show that our framework often obtains the ground truth optimal solution, and can outperform conventional methods for route optimization, which do not leverage full predictive distributions.

stat.ML↗