arXiv subjects

Xuanqi Huang

Publications and source records attributed to Xuanqi Huang.

8 recordsLinked to original sources

arXiv2026

TriALS: Triphasic-Aided Liver Lesion Segmentation Benchmark in Non-Contrast CT

Automated segmentation of liver lesions on non-contrast computed tomography (NCCT) is clinically important but fundamentally challenging, particularly in low-resource settings across Africa and Asia where contrast agents are frequently unavailable. Progress has been limited by the absence of annotated NCCT benchmarks. Here we describe the TriALS challenge for automated liver lesion segmentation under contrast-limited conditions, supported by a multi-centre dataset of 150 cases with four-phase CT acquisitions (600 volumes) from Egyptian and Chinese institutions. Algorithms were evaluated on 70 cases from three institutions, including an independent external cohort. The top-performing method achieved a mean venous-phase Dice of 0.754, consistent with human-level performance, yet dropped to 0.57 on NCCT. On external validation, the leading method outperformed off-the-shelf models by up to 28% in Dice on NCCT. Algorithm performance was most strongly predicted by training data scale and pre-training strategy. A cross-year comparison exposed a persistent perceptual barrier on NCCT that scaling pre-training alone cannot overcome. Data, annotations, and code are available at https://github.com/xmed-lab/TriALS.

cs.CV↗

arXiv2026

Distribution-Aware Reward Estimation for Test-Time Reinforcement Learning

Test-time reinforcement learning (TTRL) enables large language models (LLMs) to self-improve on unlabeled inputs, but its effectiveness critically depends on how reward signals are estimated without ground-truth supervision. Most existing TTRL methods rely on majority voting (MV) over rollouts to produce deterministic rewards, implicitly assuming that the majority rollout provides a reliable learning signal. We show that this assumption is fragile: MV reduces the rollout distribution into a single outcome, discarding information about non-majority but correct actions candidates, and yields systematically biased reward estimates. To address this, we propose Distribution-AwareReward Estimation (DARE), which shifts reward estimation from a single majority outcome to the full empirical rollout distribution. DARE further augments this distribution-based reward with an exploration bonus and a distribution pruning mechanism for non-majority rollout exploration and reward denoise, yielding a more informative and robust reward estimation. Extensive experiments on challenging reasoning benchmarks show that DARE improves optimization stability and final performance over recent baselines, achieving relative improvements of 25.3% on challenging AIME 2024 and 5.3% on AMC.

cs.CL↗

arXiv2025

Neurons: Emulating the Human Visual Cortex Improves Fidelity and Interpretability in fMRI-to-Video Reconstruction

Decoding visual stimuli from neural activity is essential for understanding the human brain. While fMRI methods have successfully reconstructed static images, fMRI-to-video reconstruction faces challenges due to the need for capturing spatiotemporal dynamics like motion and scene transitions. Recent approaches have improved semantic and perceptual alignment but struggle to integrate coarse fMRI data with detailed visual features. Inspired by the hierarchical organization of the visual system, we propose NEURONS, a novel framework that decouples learning into four correlated sub-tasks: key object segmentation, concept recognition, scene description, and blurry video reconstruction. This approach simulates the visual cortex's functional specialization, allowing the model to capture diverse video content. In the inference stage, NEURONS generates robust conditioning signals for a pre-trained text-to-video diffusion model to reconstruct the videos. Extensive experiments demonstrate that NEURONS outperforms state-of-the-art baselines, achieving solid improvements in video consistency (26.6%) and semantic-level accuracy (19.1%). Notably, NEURONS shows a strong functional correlation with the visual cortex, highlighting its potential for brain-computer interfaces and clinical applications. Code and model weights are available at: https://github.com/xmed-lab/NEURONS.

cs.CV↗

arXiv2019

Demonstration of beta-Ga2O3 Optical Waveguides and the Analysis of Their Propagation Losses in the UV-Visible Spectra

This paper reports the first demonstration of beta-phase gallium oxide as optical waveguides on sapphire substrates grown by metal-organic chemical vapor deposition (MOCVD). The propagation losses from visible to ultraviolet spectra were comprehensively studied. By optimizing the fabrication processes, minimum propagation loss was identified to be 3.7 dB/cm at the wavelength of 810 nm, which is comparable to other wide bandgap materials within the III-N family (GaN, AlN). To further reveal the underlying loss mechanisms, several physical mechanisms such as two-photon absorption, sidewall scattering, top surface scattering, and bulk scattering were taken into consideration. The results obtained from this work suggest that beta-Ga2O3 is promising for ultraviolet-visible spectrum integrated photonic applications.

physics.optics↗

arXiv2019

On-chip directional octave-spanning supercontinuum generation from high order mode in near ultraviolet to infrared spectrum using AlN waveguides

On-chip ultraviolet to infrared (UV-IR) spectrum frequency metrology is of crucial importance as a characterization tool for fundamental studies on quantum physics, chemistry, and biology. Due to the strong material dispersion, traditional techniques fail to demonstrate the device that can be applied to generate coherent broadband spectrum that covers the full UV-IR wavelengths. In this work, we explore several novel techniques for supercontinuum generation covering near-UV to near-IR spectrum using AlN micro-photonic waveguides, which is essential for frequency metrology applications: First, to create anomalous dispersion, high order mode (TE10) was adopted, together with its carefully designed high efficiency excitation strategies. Second, the spectrum was broadened by soliton fission through third order dispersion and second harmonic generation, by which directional energy transfer from near-IR to near-UV can be obtained. Finally, high quality single crystalline AlN material was used to provide broadband transparency from UV to IR. Under decently low pulse energy of 0.36 nJ, the experimental spectrum from supercontinuum generation covers from 490 nm to over 1100 nm, with a second harmonic generation band covering from 405 nm to 425 nm. This work paves the way towards UV-IR full spectrum on-chip frequency metrology applications.

physics.optics↗

arXiv2018

A Novel PN junction between Mechanically Exfoliated \b{eta}-Ga2O3 and p-GaN

Several pn junctions were constructed from mechanically exfoliated ultrawide bandgap (UWBG) beta-phase gallium oxide (\b{eta}-Ga2O3) and p-type gallium nitride (GaN). The mechanical exfoliation process, which is described in detail, is similar to that of graphene and other 2D materials. Atomic force microscopy (AFM) scans of the exfoliated \b{eta}-Ga2O3 flakes show very smooth surfaces with average roughness of 0.647 nm and transmission electron microscopy (TEM) scans reveal flat, clean interfaces between the \b{eta}-Ga2O3 flakes and p-GaN. The device showed a rectification ratio around 541.3 (V+5/V-5). Diode performance improved over the temperature range of 25°C and 200°C, leading to an unintentional donor activation energy of 135 meV. As the thickness of exfoliated \b{eta}-Ga2O3 increases, ideality factors decrease as do the diode turn on voltages, tending toward an ideal threshold voltage of 3.2 V as determined by simulation. This investigation can help increase study of novel devices between mechanically exfoliated \b{eta}-Ga2O3 and other materials.

physics.app-ph↗

arXiv2017

Effect of crystalline anisotropy on vertical (-201) and (010) beta-Ga2O3 Schottky barrier diodes on EFG single-crystal substrates

Vertical (-201) and (010) beta-Ga2O3 Schottky barrier diodes (SBDs) were fabricated on single-crystal substrates grown by edge-defined film-fed growth (EFG) method. High resolution X-ray diffraction (HRXRD) and atomic force microscopy (AFM) confirmed good crystal quality and surface morphology of the substrates. The electrical properties of both devices, including current-voltage (I-V) and capacitance-voltage (C-V) characteristics, were comprehensively measured and compared. The (-201) and (010) SBDs exhibited on-resistances (Ron) of 0.56 and 0.77 mΩcm2, turn-on voltages (Von) of 1.0 and 1.3 V, Schottky barrier heights (SBH) of 1.05 and 1.20 eV, electron mobilities of 125 and 65 cm2/(Vs), respectively, with a high on-current of ~1.3 kA/cm2 and on/off ratio of ~109. The (010) SBD had a larger Von and SBH than (-201) SBD due to anisotropic surface properties (i.e., surface Fermi level pinning and band bending), as supported by X-ray photoelectron spectroscopy (XPS) measurements. Temperature-dependent I-V also revealed the inhomogeneous nature of the SBH in both devices, where (-201) SBD showed a more uniform SBH distribution. The homogeneous SBH was also extracted: 1.33 eV for (-201) SBD and 1.53 eV for (010) SBD. The reverse leakage current of the devices was well described by the two-step trap-assisted tunneling model and the one-dimensional variable range hopping conduction (1D-VRH) model. The (-201) SBD showed larger leakage current due to its lower SBH and smaller activation energy. These results indicate the crystalline anisotropy of beta-Ga2O3 can affect the electrical properties of vertical SBDs and should be taken into consideration when designing beta-Ga2O3 electronics.

cond-mat.mtrl-sci↗

arXiv2017

Characterizations of the Nonlinear Optical Properties for (010) and (-201) Beta-phase Gallium Oxide

We report, for the first time, the characterizations on optical nonlinearities of beta-phase gallium oxide (\b{eta}-Ga2O3), where both (010) \b{eta}-Ga2O3 and (-201) \b{eta}-Ga2O3 were examined for two-photon absorption (TPA) coefficient, Kerr refractive index, and their polarization dependence. The wavelength dependence of the TPA coefficient and Kerr refractive index was estimated using a widely used analytical model. \b{eta}-Ga2O3 exhibits a TPA coefficient of 1.2 cm/GW for (010) \b{eta}-Ga2O3 and 0.58 cm/GW for (-201) \b{eta}-Ga2O3. The Kerr refractive index is -2.14*10^(15) cm2/W for (010) \b{eta}-Ga2O3 and -2.89*10^(15) cm2/W for (-201) \b{eta}-Ga2O3. In addition, \b{eta}-Ga2O3 shows stronger nonlinear optical anisotropy on the (-201) plane than on the (010) plane, possibly due to highly asymmetric crystal structure. Compared with that of gallium nitride (GaN), the TPA coefficient of \b{eta}-Ga2O3 is 20 times smaller, and the Kerr refractive index of \b{eta}-Ga2O3 is also found to be 4 to 5 times smaller. These results indicate that \b{eta}-Ga2O3 has the potential for use in ultra-low loss waveguides and ultra-stable resonators and integrated photonics, especially in the UV and visible wavelength spectral range.

physics.app-ph↗