arXiv subjects

Aggelos K. Katsaggelos

Publications and source records attributed to Aggelos K. Katsaggelos.

At least 19 recordsLinked to original sources

arXiv2026

Vision-Language Enhanced Foundation Model for Semi-Supervised Medical Image Segmentation

Semi-supervised learning (SSL) has emerged as an efficient paradigm for medical image segmentation, reducing the reliance on extensive expert annotations. Vision-language models (VLMs) have demonstrated strong generalization and few-shot capabilities across diverse visual domains. In this work, we integrate a VLM into a semi-supervised medical image segmentation model by adding a Vision-Language Enhanced Semi-supervised Segmentation Assistant (VESSA) that incorporates foundation-level visual-semantic understanding into SSL frameworks. Our approach consists of two stages. In Stage 1, the VLM-enhanced segmentation foundation model VESSA is trained as a reference-guided segmentation assistant using a template bank containing gold-standard exemplars, simulating learning from limited labeled data. Given an input-template pair, VESSA performs visual feature matching to extract representative semantic and spatial cues from exemplar segmentations, generating structured prompts for a Segment Anything Model (SAM)-inspired mask decoder to produce segmentation masks. In Stage 2, VESSA is integrated into an SSL framework as a plug-and-play teacher, providing template-guided pseudo-labels that complement the task-specific student model and strengthen supervision under scarce annotations. Extensive experiments across multiple segmentation datasets and domains show that VESSA-augmented SSL significantly enhances segmentation accuracy, outperforming state-of-the-art baselines under extremely limited annotation conditions.

cs.CV↗

arXiv2026

MetaSR: Content-Adaptive Metadata Orchestration for Generative Super-Resolution

We study generative super-resolution (SR) in real-world scenarios where content and degradations vary across domains, genres, and segments. For example, images and videos may alternate between text overlays, fast motion, smooth cartoons, and low-light faces, each benefiting from different forms of side information. Existing metadata-guided SR methods typically use a fixed conditioning design, which is suboptimal when useful cues are content dependent and transmission budgets are limited. We propose MetaSR, a Diffusion Transformer (DiT)-based framework that selects and injects task-relevant metadata to guide SR under resource constraints. Specifically, we use the DiT's own VAE and transformer backbone to fuse heterogeneous metadata, and adopt an efficient distillation strategy that enables one-step diffusion inference. Experiments across diverse content buckets and degradation regimes show that MetaSR outperforms reference solutions by up to 1.0~dB PSNR while achieving up to 50\% transmission bitrate saving at matched quality. We assess these gains under a rate--distortion optimization (RDO) framework that jointly accounts for sender-side bitrate and receiver/display quality metrics (e.g., PSNR and SSIM).

cs.CV↗

arXiv2026

Hallucination-Free Automatic Question & Answer Generation for Intuitive Learning

Hallucinations in large language models (LLMs), defined as fluent yet incorrect or incoherent outputs, pose a significant challenge to the automatic generation of educational multiple-choice questions (MCQs). We identified four key hallucination types in MCQ generation: reasoning inconsistencies, insolvability, factual errors, and mathematical errors. To address this, we propose a hallucination-free multi-agent generation framework that breaks down MCQ generation into discrete, verifiable stages. Our framework utilizes both rule-based and LLM-based detection agents, as well as hallucination scoring metrics to optimize question quality. We redefined MCQ generation as an optimization task minimizing hallucination risk while maximizing validity, answerability, and cost-efficiency. We also introduce an agent-led refinement process that uses counterfactual reasoning and chain-of-thought (CoT) to iteratively improve hallucination in question generation. We evaluated a sample of AP- aligned STEM questions, where our system reduced hallucination rates by over 90% compared to baseline generation while preserving the educational value and style of questions. Our results demonstrate that structured multi-agent collaboration can mitigate hallucinations in educational content creation at scale, paving the way for more reliable LLM-powered learning tools.

cs.CL↗

arXiv2026

Automatic Question Generation for Intuitive Learning Utilizing Causal Graph Guided Chain of Thought Reasoning

Intuitive learning is crucial for developing deep conceptual understanding, especially in STEM education, where students often struggle with abstract and interconnected concepts. Automatic question generation has become an effective strategy for personalized and adaptive learning. However, its effectiveness is hindered by hallucinations in large language models (LLMs), which may generate factually incorrect, ambiguous, or pedagogically inconsistent questions. To address this issue, we propose a novel framework that combines causal-graph-guided Chain-of-Thought (CoT) reasoning with a multi-agent LLM architecture. This approach ensures the generation of accurate, meaningful, and curriculum-aligned questions. Causal graphs provide an explicit representation of domain knowledge, while CoT reasoning facilitates a structured, step-by-step traversal of related concepts. Dedicated LLM agents are assigned specific tasks such as graph pathfinding, reasoning, validation, and output, all working within domain constraints. A dual validation mechanism-at both the conceptual and output stages-greatly reduces hallucinations. Experimental results demonstrate up to a 70% improvement in quality compared to reference methods and yielded highly favorable outcomes in subjective evaluations.

cs.AI↗

arXiv2025

Physics-Informed Image Restoration via Progressive PDE Integration

Motion blur, caused by relative movement between camera and scene during exposure, significantly degrades image quality and impairs downstream computer vision tasks such as object detection, tracking, and recognition in dynamic environments. While deep learning-based motion deblurring methods have achieved remarkable progress, existing approaches face fundamental challenges in capturing the long-range spatial dependencies inherent in motion blur patterns. Traditional convolutional methods rely on limited receptive fields and require extremely deep networks to model global spatial relationships. These limitations motivate the need for alternative approaches that incorporate physical priors to guide feature evolution during restoration. In this paper, we propose a progressive training framework that integrates physics-informed PDE dynamics into state-of-the-art restoration architectures. By leveraging advection-diffusion equations to model feature evolution, our approach naturally captures the directional flow characteristics of motion blur while enabling principled global spatial modeling. Our PDE-enhanced deblurring models achieve superior restoration quality with minimal overhead, adding only approximately 1\% to inference GMACs while providing consistent improvements in perceptual quality across multiple state-of-the-art architectures. Comprehensive experiments on standard motion deblurring benchmarks demonstrate that our physics-informed approach improves PSNR and SSIM significantly across four diverse architectures, including FFTformer, NAFNet, Restormer, and Stripformer. These results validate that incorporating mathematical physics principles through PDE-based global layers can enhance deep learning-based image restoration, establishing a promising direction for physics-informed neural network design in computer vision applications.

cs.CV↗

arXiv2025

Advancing Limited-Angle CT Reconstruction Through Diffusion-Based Sinogram Completion

Limited Angle Computed Tomography (LACT) often faces significant challenges due to missing angular information. Unlike previous methods that operate in the image domain, we propose a new method that focuses on sinogram inpainting. We leverage MR-SDEs, a variant of diffusion models that characterize the diffusion process with mean-reverting stochastic differential equations, to fill in missing angular data at the projection level. Furthermore, by combining distillation with constraining the output of the model using the pseudo-inverse of the inpainting matrix, the diffusion process is accelerated and done in a step, enabling efficient and accurate sinogram completion. A subsequent post-processing module back-projects the inpainted sinogram into the image domain and further refines the reconstruction, effectively suppressing artifacts while preserving critical structural details. Quantitative experimental results demonstrate that the proposed method achieves state-of-the-art performance in both perceptual and fidelity quality, offering a promising solution for LACT reconstruction in scientific and clinical applications.

cs.CV↗

arXiv2025

RN-SDEs: Limited-Angle CT Reconstruction with Residual Null-Space Diffusion Stochastic Differential Equations

Computed tomography is a widely used imaging modality with applications ranging from medical imaging to material analysis. One major challenge arises from the lack of scanning information at certain angles, resulting in distortion or artifacts in the reconstructed images. This is referred to as the Limited Angle Computed Tomography (LACT) reconstruction problem. To address this problem, we propose the use of Residual Null-Space Diffusion Stochastic Differential Equations (RN-SDEs), which are a variant of diffusion models that characterize the diffusion process with mean-reverting (MR) stochastic differential equations. To demonstrate the generalizability of RN-SDEs, we conducted experiments with two different LACT datasets, ChromSTEM and C4KC-KiTS. Through extensive experiments, we demonstrate that by leveraging learned MR-SDEs as a prior and emphasizing data consistency using Range-Null Space Decomposition (RNSD) based rectification, we can recover high-quality images from severely degraded ones and achieve state-of-the-art performance in most LACT tasks. Additionally, we present a quantitative comparison of RN-SDE with other networks, in terms of computational complexity and runtime efficiency, highlighting the superior effectiveness of our proposed approach.

eess.IV↗

arXiv2025

Caption-Driven Explainability: Probing CNNs for Bias via CLIP

Robustness has become one of the most critical problems in machine learning (ML). The science of interpreting ML models to understand their behavior and improve their robustness is referred to as explainable artificial intelligence (XAI). One of the state-of-the-art XAI methods for computer vision problems is to generate saliency maps. A saliency map highlights the pixel space of an image that excites the ML model the most. However, this property could be misleading if spurious and salient features are present in overlapping pixel spaces. In this paper, we propose a caption-based XAI method, which integrates a standalone model to be explained into the contrastive language-image pre-training (CLIP) model using a novel network surgery approach. The resulting caption-based XAI model identifies the dominant concept that contributes the most to the models prediction. This explanation minimizes the risk of the standalone model falling for a covariate shift and contributes significantly towards developing robust ML models. Our code is available at https://github.com/patch0816/caption-driven-xai

cs.CV↗

arXiv2025

Advancing Glitch Classification in Gravity Spy: Multi-view Fusion with Attention-based Machine Learning for Advanced LIGO's Fourth Observing Run

The first successful detection of gravitational waves by ground-based observatories, such as the Laser Interferometer Gravitational-Wave Observatory (LIGO), marked a breakthrough in our comprehension of the Universe. However, due to the unprecedented sensitivity required to make such observations, gravitational-wave detectors also capture disruptive noise sources called glitches, which can potentially be confused for or mask gravitational-wave signals. To address this problem, a community-science project, Gravity Spy, incorporates human insight and machine learning to classify glitches in LIGO data. The machine-learning classifier, integrated into the project since 2017, has evolved over time to accommodate increasing numbers of glitch classes. Despite its success, limitations have arisen in the ongoing LIGO fourth observing run (O4) due to the architecture's simplicity, which led to poor generalization and inability to handle multi-time window inputs effectively. We propose an advanced classifier for O4 glitches. Using data from previous observing runs, we evaluate different fusion strategies for multi-time window inputs, using label smoothing to counter noisy labels, and enhancing interpretability through attention module-generated weights. Our new O4 classifier shows improved performance, and will enhance glitch classification, aiding in the ongoing exploration of gravitational-wave phenomena.

gr-qc↗

arXiv2025

Probabilistic smooth attention for deep multiple instance learning in medical imaging

The Multiple Instance Learning (MIL) paradigm is attracting plenty of attention in medical imaging classification, where labeled data is scarce. MIL methods cast medical images as bags of instances (e.g. patches in whole slide images, or slices in CT scans), and only bag labels are required for training. Deep MIL approaches have obtained promising results by aggregating instance-level representations via an attention mechanism to compute the bag-level prediction. These methods typically capture both local interactions among adjacent instances and global, long-range dependencies through various mechanisms. However, they treat attention values deterministically, potentially overlooking uncertainty in the contribution of individual instances. In this work we propose a novel probabilistic framework that estimates a probability distribution over the attention values, and accounts for both global and local interactions. In a comprehensive evaluation involving {\color{review} eleven} state-of-the-art baselines and three medical datasets, we show that our approach achieves top predictive performance in different metrics. Moreover, the probabilistic treatment of the attention provides uncertainty maps that are interpretable in terms of illness localization.

cs.CV↗

arXiv2025

Harmony-Aware Music-driven Motion Synthesis with Perceptual Constraint on UGC Datasets

With the popularity of video-based user-generated content (UGC) on social media, harmony, as dictated by human perceptual principles, is critical in assessing the rhythmic consistency of audio-visual UGCs for better user engagement. In this work, we propose a novel harmony-aware GAN framework, following a specifically designed harmony evaluation strategy to enhance rhythmic synchronization in the automatic music-to-motion synthesis using a UGC dance dataset. This harmony strategy utilizes refined cross-modal beat detection to capture closely correlated audio and visual rhythms in an audio-visual pair. To mimic human attention mechanism, we introduce saliency-based beat weighting and interval-driven beat alignment, which ensures accurate harmony score estimation consistent with human perception. Building on this strategy, our model, employing efficient encoder-decoder and depth-lifting designs, is adversarially trained based on categorized musical meter segments to generate realistic and rhythmic 3D human motions. We further incorporate our harmony evaluation strategy as a weakly supervised perceptual constraint to flexibly guide the synchronized audio-visual rhythms during the generation process. Experimental results show that our proposed model significantly outperforms other leading music-to-motion methods in rhythmic harmony, both quantitatively and qualitatively, even with limited UGC training data. Live samples 15 can be watched at: https://youtu.be/tWwz7yq4aUs

cs.MM↗

arXiv2025

Efficient Lung Ultrasound Severity Scoring Using Dedicated Feature Extractor

With the advent of the COVID-19 pandemic, ultrasound imaging has emerged as a promising technique for COVID-19 detection, due to its non-invasive nature, affordability, and portability. In response, researchers have focused on developing AI-based scoring systems to provide real-time diagnostic support. However, the limited size and lack of proper annotation in publicly available ultrasound datasets pose significant challenges for training a robust AI model. This paper proposes MeDiVLAD, a novel pipeline to address the above issue for multi-level lung-ultrasound (LUS) severity scoring. In particular, we leverage self-knowledge distillation to pretrain a vision transformer (ViT) without label and aggregate frame-level features via dual-level VLAD aggregation. We show that with minimal finetuning, MeDiVLAD outperforms conventional fully-supervised methods in both frame- and video-level scoring, while offering classification reasoning with exceptional quality. This superior performance enables key applications such as the automatic identification of critical lung pathology areas and provides a robust solution for broader medical video classification tasks.

eess.IV↗

arXiv2025

ScarNet: A Novel Foundation Model for Automated Myocardial Scar Quantification from LGE in Cardiac MRI

Background: Late Gadolinium Enhancement (LGE) imaging is the gold standard for assessing myocardial fibrosis and scarring, with left ventricular (LV) LGE extent predicting major adverse cardiac events (MACE). Despite its importance, routine LGE-based LV scar quantification is hindered by labor-intensive manual segmentation and inter-observer variability. Methods: We propose ScarNet, a hybrid model combining a transformer-based encoder from the Medical Segment Anything Model (MedSAM) with a convolution-based U-Net decoder, enhanced by tailored attention blocks. ScarNet was trained on 552 ischemic cardiomyopathy patients with expert segmentations of myocardial and scar boundaries and tested on 184 separate patients. Results: ScarNet achieved robust scar segmentation in 184 test patients, yielding a median Dice score of 0.912 (IQR: 0.863--0.944), significantly outperforming MedSAM (median Dice = 0.046, IQR: 0.043--0.047) and nnU-Net (median Dice = 0.638, IQR: 0.604--0.661). ScarNet demonstrated lower bias (-0.63%) and coefficient of variation (4.3%) compared to MedSAM (bias: -13.31%, CoV: 130.3%) and nnU-Net (bias: -2.46%, CoV: 20.3%). In Monte Carlo simulations with noise perturbations, ScarNet achieved significantly higher scar Dice (0.892 \pm 0.053, CoV = 5.9%) than MedSAM (0.048 \pm 0.112, CoV = 233.3%) and nnU-Net (0.615 \pm 0.537, CoV = 28.7%). Conclusion: ScarNet outperformed MedSAM and nnU-Net in accurately segmenting myocardial and scar boundaries in LGE images. The model exhibited robust performance across diverse image qualities and scar patterns.

eess.IV↗

arXiv2024

Sm: enhanced localization in Multiple Instance Learning for medical imaging classification

Multiple Instance Learning (MIL) is widely used in medical imaging classification to reduce the labeling effort. While only bag labels are available for training, one typically seeks predictions at both bag and instance levels (classification and localization tasks, respectively). Early MIL methods treated the instances in a bag independently. Recent methods account for global and local dependencies among instances. Although they have yielded excellent results in classification, their performance in terms of localization is comparatively limited. We argue that these models have been designed to target the classification task, while implications at the instance level have not been deeply investigated. Motivated by a simple observation -- that neighboring instances are likely to have the same label -- we propose a novel, principled, and flexible mechanism to model local dependencies. It can be used alone or combined with any mechanism to model global dependencies (e.g., transformers). A thorough empirical validation shows that our module leads to state-of-the-art performance in localization while being competitive or superior in classification. Our code is at https://github.com/Franblueee/SmMIL.

cs.CV↗

arXiv2024

Longitudinal Wrist PPG Analysis for Reliable Hypertension Risk Screening Using Deep Learning

Hypertension is a leading risk factor for cardiovascular diseases. Traditional blood pressure monitoring methods are cumbersome and inadequate for continuous tracking, prompting the development of PPG-based cuffless blood pressure monitoring wearables. This study leverages deep learning models, including ResNet and Transformer, to analyze wrist PPG data collected with a smartwatch for efficient hypertension risk screening, eliminating the need for handcrafted PPG features. Using the Home Blood Pressure Monitoring (HBPM) longitudinal dataset of 448 subjects and five-fold cross-validation, our model was trained on over 68k spot-check instances from 358 subjects and tested on real-world continuous recordings of 90 subjects. The compact ResNet model with 0.124M parameters performed significantly better than traditional machine learning methods, demonstrating its effectiveness in distinguishing between healthy and abnormal cases in real-world scenarios.

eess.SP↗

arXiv2024

DRL-STNet: Unsupervised Domain Adaptation for Cross-modality Medical Image Segmentation via Disentangled Representation Learning

Unsupervised domain adaptation (UDA) is essential for medical image segmentation, especially in cross-modality data scenarios. UDA aims to transfer knowledge from a labeled source domain to an unlabeled target domain, thereby reducing the dependency on extensive manual annotations. This paper presents DRL-STNet, a novel framework for cross-modality medical image segmentation that leverages generative adversarial networks (GANs), disentangled representation learning (DRL), and self-training (ST). Our method leverages DRL within a GAN to translate images from the source to the target modality. Then, the segmentation model is initially trained with these translated images and corresponding source labels and then fine-tuned iteratively using a combination of synthetic and real images with pseudo-labels and real labels. The proposed framework exhibits superior performance in abdominal organ segmentation on the FLARE challenge dataset, surpassing state-of-the-art methods by 11.4% in the Dice similarity coefficient and by 13.1% in the Normalized Surface Dice metric, achieving scores of 74.21% and 80.69%, respectively. The average running time is 41 seconds, and the area under the GPU memory-time curve is 11,292 MB. These results indicate the potential of DRL-STNet for enhancing cross-modality medical image segmentation tasks.

eess.IV↗

arXiv2024

Characterization of Crystal Properties and Defects in CdZnTe Radiation Detectors

CdZnTe-based detectors are highly valued because of their high spectral resolution, which is an essential feature for nuclear medical imaging. However, this resolution is compromised when there are substantial defects in the CdZnTe crystals. In this study, we present a learning-based approach to determine the spatially dependent bulk properties and defects in semiconductor detectors. This characterization allows us to mitigate and compensate for the undesired effects caused by crystal impurities. We tested our model with computer-generated noise-free input data, where it showed excellent accuracy, achieving an average RMSE of 0.43% between the predicted and the ground truth crystal properties. In addition, a sensitivity analysis was performed to determine the effect of noisy data on the accuracy of the model.

cond-mat.mtrl-sci↗

arXiv2024

Review and Novel Formulae for Transmittance and Reflectance of Wedged Thin Films on absorbing Substrates

Historically, spectroscopic techniques have been essential for studying the optical properties of thin solid films. However, existing formulae for both normal transmission and reflection spectroscopy often rely on simplified theoretical assumptions, which may not accurately align with real-world conditions. For instance, it is common to assume (1) that the thin solid layers are deposited on completely transparent thick substrates and (2) that the film surface forms a specular plane with a relatively small wedge angle. While recent studies have addressed these assumptions separately, this work presents an integrated framework that eliminates both assumptions simultaneously. In addition, the current work presents a deep review of various formulae from the literature, each with their corresponding levels of complexity. Our review analysis highlights a critical trade-off between computational complexity and expression accuracy, where the newly developed formulae offer enhanced accuracy at the expense of increased computational time. Our user-friendly code, which includes several classical transmittance and reflectance formulae from the literature and our newly proposed expressions, is publicly available in both Python and Matlab at this link.

cond-mat.mtrl-sci↗