Underwater images captured in real marine environments usually suffer from wavelength-dependent color attenuation, low contrast, haze-like scattering, and weak structural details. Although supervised underwater image enhancement (UIE) methods have achieved promising results, their performance is still strongly affected by the availability and reliability of paired underwater training data. In addition, commonly used pixel-level or feature-level losses mainly constrain the image modality itself and provide limited guidance from high-level perceptual semantics. To address these problems, this paper proposes PhyCR-UIE, an unsupervised underwater image enhancement framework that integrates physical imaging priors and cross-modal perceptual regularization. Specifically, three CNN-based subnetworks are designed to estimate the global background light, the medium transmission map, and the latent scene radiance in the underwater imaging model. The estimated physical variables are jointly optimized by reconstruction consistency and homology consistency, enabling the network to learn from degraded underwater images without paired clean references. Furthermore, a cross-modal loss based on CLIP-based image quality assessment (CLIP-IQA+) is introduced. By measuring the semantic similarity between the predicted scene radiance and positive and negative image-quality descriptions, this loss guides the enhanced result toward clearer, more natural, and color-balanced visual appearances. Experiments on five real underwater datasets demonstrate favorable color correction, perceptual quality, and underwater image quality ranking.
Photoacoustic (PA) imaging enables high-contrast visualization of blood vessels; however, accurate 3-D vascular imaging requires volumetric information across a wide frequency band. Sparse-element 2-D matrix array transducers can enable rapid and cost-effective 3-D imaging. However, when used with conventional universal back projection (UBP) techniques, these systems often suffer from image degradation. Model-based learning (MBLr), which integrates physical models with deep learning, has emerged as a promising approach to address this limitation and improve image quality from sparse sensor configurations. In this study, we performed simulation-based analyses of 3-D PA images to investigate the extent to which MBLr can achieve high contrast using a sparse, cost-effective matrix array and to examine the mechanism underlying its image quality improvements. Quantitative image quality metrics showed that MBLr provided significant improvements compared with conventional UBP reconstruction methods. Specifically, a sparse 16 × 16 array (Case S) reconstructed using MBLr outperformed both the baseline 32 × 32 array (Case B) and the ultra-wideband 32 × 32 array (Case U). Spatial frequency analysis revealed that MBLr enhanced high-frequency recovery in the in-plane (x and y) directions and improved low-frequency components in the axial (z) direction. These improvements enabled substantially enhanced 3-D vascular visualization with superior vessel delineation, contrast, and structural preservation. Overall, our findings demonstrate that MBLr enables high-quality volumetric imaging with sparse array transducer, providing important insights for the design of cost-effective PA imaging systems.
Virtual reality (VR)-based evacuation training provides a safe and immersive environment for participants to experience disaster scenarios. However, existing systems often prioritize the experience itself, leaving the critical stage of reflection-essential for refining and stabilizing evacuation knowledge-under-supported. This study presents a qualitative pilot investigation into an extended reflection support function for a VR-based evacuation training system. Unlike traditional replay functions that only visualize avatar movements, our system synchronizes spatialized environmental sounds and recorded verbal utterances, i.e., voices of the user and non-player characters (NPCs), with the visual replay. A preliminary experiment involving eight university students was conducted to evaluate how these auditory clues influence the reflection-on-action process. Qualitative results indicate that audio clues help participants recall their internal decision-making processes and provide essential context for understanding the actions of others (NPCs). The findings suggest that the integration of auditory information facilitates evacuation knowledge refinement, i.e., the transition from mere experience to the formulation of concrete survival concepts. Although limited by a small sample size, this study highlights the potential of multi-modal reflection support in VR-based evacuation training.
An uncommon power-generation phenomenon in droplet-driven electricity generation on high-quality epitaxial graphene films grown on SiC substrates was investigated. This phenomenon cannot be accounted for using conventional models. The measured voltage response was isotropic and independent of the droplet sliding direction and electrode geometry. In addition, electricity generation was independent of the graphene area and flow distance. The observed nonmonotonic voltage-ion concentration relationship, featuring distinct concentration-dependent peaks, together with the independence of the load resistance, is consistent with a microscopic, discontinuous generation process. These results highlight the role of nanoscale interfacial dynamics in the hydrovoltaic effect and indicate that atomic-scale control of the graphene substrate structure may inform the design of future energy harvesters.
Nivolumab plus ipilimumab (NIVO + IPI) is a standard first-line therapy for intermediate-risk and poor-risk metastatic renal cell carcinoma (mRCC), but the long-term prognostic impact of prior primary tumor resection in this setting remains unclear. We retrospectively reviewed patients with mRCC that received NIVO + IPI as first-line therapy at eight Japanese institutions between October 2015 and May 2022. Patients were stratified into groups according to timing of metastasis and prior primary tumor resection (cytoreductive nephrectomy; CN): a metachronous group, a synchronous/CN(+) group, and a synchronous/CN(−) group. Progression-free survival, overall survival, and tumor responses were compared among groups, with a median follow-up of 54 months. Among 135 eligible patients, 40 were classified into the metachronous group (30
