Search Results (22)

Background: Current artificial intelligence (AI) systems in gastrointestinal (GI) endoscopy primarily emphasize binary detection or static classification, providing limited support for the graded assessment of malignant potential that underpins clinical decision-making. We developed GastroMalign, a transformer-based framework designed to stratify GI lesions according to ordinal disease severity while maintaining clinical interpretability, addressing this unmet need in endoscopic risk assessment. Methods: This retrospective development and validation study used the publicly available GastroVision dataset, comprising 8000 de-identified endoscopic still images from the upper and lower gastrointestinal tract, including the esophagus, stomach, duodenum, colon, rectum, and terminal ileum. GastroMalign integrates a Vision Transformer (ViT) encoder with a Sequential Feature Learner that explicitly models ordinal disease severity along a benign-to-malignant spectrum. The framework produces both categorical risk classification and a continuous malignancy risk score. Images were stratified into training (80%), validation (10%), and test (10%) sets. Performance was compared with convolutional neural network (CNN) baselines and a Swin Transformer. Interpretability was assessed using Score-CAM visualizations reviewed by blinded expert endoscopists. Results: On the held-out test set (n = 800 images), GastroMalign achieved an overall accuracy of 80.06%, precision of 79.65%, recall of 80.06%, and F1-score of 79.17%, with a micro-averaged AUC of 0.98. In comparison, ResNet-50 and DenseNet-121 achieved accuracies of 32.42% and 36.77%, respectively, while the Swin Transformer achieved 60.56% accuracy (AUC = 0.93). Ablation analyses demonstrated a 17% absolute reduction in High-Risk lesion recall when the progression-aware module was removed. Continuous malignancy risk scores increased monotonically across ordinal classes, with mean values < 0.18 for Benign and >0.72 for High-Risk/Malignant lesions. Score-CAM visualizations demonstrated 92% overlap with clinician-annotated lesion regions. Conclusions: GastroMalign delivers an interpretable, progression-aware AI framework for GI lesion risk stratification that outperforms existing CNN- and transformer-based models. Clinically, GastroMalign is intended as an adjunct decision-support tool during endoscopic review to standardize lesion risk stratification (benign to malignant spectrum), support management decisions (biopsy vs. resection vs. surveillance), and reduce operator-dependent variability by pairing ordinal risk outputs with interpretable visual explanations. Full article

Background: Primary open-angle glaucoma (POAG) is a leading cause of irreversible blindness worldwide, characterized by progressive optic nerve degeneration and marked molecular heterogeneity. Increasing evidence indicates that metabolic dysregulation and immune remodeling contribute to POAG pathogenesis; however, the underlying regulatory networks and reliable diagnostic biomarkers remain incompletely defined. Methods: Bulk transcriptomic and single-cell RNA sequencing (scRNA-seq) datasets of trabecular meshwork tissues were retrieved from Gene Expression Omnibus (GEO). Differential expression analysis and weighted gene co-expression network analysis (WGCNA) were performed to identify disease-associated modules. A machine learning framework was applied to construct classification models. Estimated immune-cell fractions were assessed using CIBERSORT, followed by pathway and transcription factor analyses. Single-cell analysis was conducted to examine the cell type-specific expression patterns. Results: A total of 195 differentially expressed genes were identified between POAG and control samples. WGCNA revealed a lactylation-related module strongly correlated with disease status. Machine learning identified S100A2 and S100A14 as candidate diagnostic biomarkers with consistent classification performance across datasets. Immune infiltration analysis suggested alterations in the immune microenvironment in POAG. Single-cell data showed that the model genes exhibited sparse but non-uniform expression across cell populations. Conclusions: This integrative analysis prioritizes S100A2 and S100A14 as candidate diagnostic biomarkers for POAG and indicates potential associations with immune-metabolic regulatory mechanisms. Full article

Background/Objectives: Glaucoma is a leading cause of irreversible blindness worldwide, making accurate and efficient detection methods essential. One primary concern with glaucoma is that it often presents no early symptoms. Vision loss typically begins at the periphery and progresses unnoticed until it significantly affects central vision. Due to this gradual and usually silent progression, early detection through regular eye exams is vital for preventing permanent vision loss. Methods: In this study, we propose a hybrid attention mechanism that recalibrates feature maps from the feature extractor for glaucoma detection. We explored normalization-free ResNet (NF-ResNet) architectures to evaluate the proposed attention mechanism, specifically NF-ResNet-26, NF-ResNet-50, and NF-ResNet-101, in comparison to baseline state-of-the-art ResNet variants. Our approach was evaluated on three publicly available glaucoma datasets, LAG, EyePACS, and BrG, to differentiate between normal and glaucomatous from fundus images. Results: The experimental results demonstrate that our proposed hybrid attention module, combined with normalization-free architectures, significantly enhances performance compared to state-of-the-art ResNet variants. The proposed attention model based on the normalization-free ResNet-50 achieved an accuracy of 0.9394 on the LAG dataset, 0.9117 on the EyePACS dataset, and 0.9020 on the BrG dataset. When evaluated on the combined dataset, the model achieved an accuracy of 0.9193, sensitivity of 0.9182, and specificity of 0.9202. Conclusions: The results from these representative datasets for glaucoma detection highlight the exceptional performance of our attention module, establishing it as a highly competitive classification model in the field of glaucoma detection. Full article

Background: Age-related macular degeneration (AMD) is a major cause of vision loss, yet publicly available Optical Coherence Tomography (OCT) datasets lack demographic diversity, particularly from South Asian populations. Existing datasets largely represent Western cohorts, limiting AI generalizability. Moreover, raw OCT volumes contain redundant spatial information and speckle noise, hindering efficient analysis. Methods: We introduce BanglaOCT2025, a retrospective dataset collected from the National Institute of Ophthalmology and Hospital (NIOH), Bangladesh, using Nidek RS-330 Duo 2 and RS-3000 Advance systems. We propose a novel preprocessing pipeline comprising two stages: (1) A constraint-based centroid minimization algorithm automatically localizes the foveal center and extracts a fixed 33-slice macular sub-volume, robust to retinal tilt and acquisition variability; and (2) A self-supervised volumetric denoising module based on a Flip-Flop Swin Transformer (FFSwin) backbone suppresses speckle noise without requiring paired clean reference data. Results: The dataset comprises 1585 OCT volumes (202,880 B-scans), including 857 expert-annotated cases (54 DryAMD, 61 WetAMD, and 742 NonAMD). Denoising quality was evaluated using reference-free volumetric metrics, paired statistical analysis, and blinded clinical review by a retinal specialist, confirming preservation of pathological biomarkers and absence of hallucination. Under a controlled paired evaluation using the same classifier with frozen weights, downstream AMD classification accuracy improved from 69.08% to 99.88%, interpreted as an upper-bound estimate of diagnostic signal recoverability rather than independent generalization. Conclusions: BanglaOCT2025 is the first clinically validated OCT dataset representing the Bengali population and establishes a reproducible fovea-centric volumetric preprocessing and restoration framework for AMD analysis, with future validation across independent and multi-centre test cohorts. Full article

With increasingly scarce spectrum resources, frequency-domain signal overlap interference has become a critical issue, making multi-user modulation classification (MUMC) a significant challenge in wireless communications. Unlike single-user modulation classification (SUMC), MUMC suffers from feature degradation caused by signal aliasing, feature redundancy, and low inter-class discriminability. To address these challenges, this paper proposes a collaborative “separation–recognition” framework. The framework begins by separating overlapping signals via a band partitioning and FastICA module to alleviate feature degradation. For the recognition phase, we design a dual-branch network: one branch extracts prior knowledge features, including amplitude, phase, and frequency, from the I/Q sequence and models their temporal dependencies using a bidirectional LSTM; the other branch learns deep hierarchical representations directly from the raw signal through multi-scale convolutional layers. The features from both branches are then adaptively fused using a gated fusion module. Experimental results show that the proposed method achieves superior performance over several baseline models across various signal conditions, validating the efficacy of the dual-branch architecture and the overall framework. Full article

Improving the decoding accuracy of biological signals has been a research focus for decades to advance health, automation, and robotic industries. However, challenges like inter-subject variability, data scarcity, and multifunctional variability cause low decoding accuracy, thus hindering the practical deployment of biological signal paradigms. This paper proposes a multifunctional biological signals network (Multi-BioSig-Net) that addresses the aforementioned issues by devising a novel blind few-shot learning (FSL) technique to quickly adapt to multiple target domains without needing a pre-trained model. Specifically, our proposed multimodal similarity extractor (MMSE) and self-multiple domain adaptation (SMDA) modules address data scarcity and inter-subject variability issues by exploiting and enhancing the similarity between multimodal samples and quickly adapting the target domains by adaptively adjusting the parameters’ weights and position, respectively. For multifunctional learning, we proposed inter-function discriminator (IFD) that discriminates the classes by extracting inter-class common features and then subtracts them from both classes to avoid false prediction of the proposed model due to overfitting on the common features. Furthermore, we proposed a holistic-local fusion (HLF) module that exploits contextual-detailed features to adapt the scale-varying features across multiple functions. In addition, fractal dimension estimation (FDE) was employed for the classification of left-hand motor imagery (LMI) and right-hand motor imagery (RMI), confirming that proposed method can effectively extract the discriminative features for this task. The effectiveness of our proposed algorithm was assessed quantitatively and statistically against competent state-of-the-art (SOTA) algorithms utilizing three public datasets, demonstrating that our proposed algorithm outperformed SOTA algorithms. Full article

Retinal diseases account for a large fraction of global blinding disorders, requiring sophisticated diagnostic tools for early management. In this study, the author proposes a hybrid deep learning framework in the form of AdaptiveSwin-CNN that combines Swin Transformers and Convolutional Neural Networks (CNNs) for the classification of multi-class retinal diseases. In contrast to traditional architectures, AdaptiveSwin-CNN utilizes a brand-new Self-Attention Fusion Module (SAFM) to effectively combine multi-scale spatial and contextual options to alleviate class imbalance and give attention to refined retina lesions. Utilizing the adaptive baseline augmentation and dataset-driven preprocessing of input images, the AdaptiveSwin-CNN model resolves the problem of the variability of fundus images in the dataset. AdaptiveSwin-CNN achieved a mean accuracy of 98.89%, sensitivity of 95.2%, specificity of 96.7%, and F1-score of 97.2% on RFMiD and ODIR benchmarks, outperforming other solutions. An additional lightweight ensemble XGBoost classifier to reduce overfitting and increase interpretability also increased diagnostic accuracy. The results highlight AdaptiveSwin-CNN as a robust and computationally efficient decision-support system. Full article

With the population aging globally, cataracts have become one of the main causes of vision impairment. Early diagnosis and treatment of cataracts are crucial for preventing blindness. However, the use of deep learning models for assisting in the diagnosis of cataracts is limited due to reasons such as scarce data labeling, small sample size, uneven distribution, and poor generalization ability in the field. Therefore, this paper proposes a hybrid deep learning network for assisting in the diagnosis of cataract fundus images, attempting to solve the above problems and limitations. The network is based on the idea of transfer learning for feature extraction of fundus images, and introduces the Squeeze-and-Excitation (SE) module and prototype network for feature enhancement and classification, improving the model’s generalization ability for new disease samples. Finally, this paper verifies the role of each part of the model through ablation experiments, especially the significant contribution of the SE_block module and the prototype network classifier in enhancing the model’s performance. The experimental results show that the proposed model achieves excellent performance in the task of cataract fundus image recognition, with an accuracy of 0.9875, AUC value of 0.9984, and F1 score of 0.9855. The establishment of this hybrid model not only provides an effective tool for the auxiliary diagnosis of cataracts but also provides a new perspective and method for the application of deep learning in the field of ophthalmic disease recognition. Full article

The computer-assisted inverse design of photonic computing, especially by leveraging artificial intelligence algorithms, offers great convenience to accelerate the speed of development and improve calculation accuracy. However, traditional thickness-based modulation methods are hindered by large volume and difficult fabrication process, making it hard to meet the data-driven requirements of flexible light modulation. Here, we propose a diffractive deep neural network (D²NN) framework based on a three-layer all-dielectric phased transmitarray as hidden layers, which can perform the classification of handwritten digits. By tailoring the radius of a silicon nanodisk of a meta-atom, the metasurface can realize the phase profile calculated by D²NN and maintain a relative high transmittance of 0.9 at a wavelength of 600 nm. The designed image classifier consists of three layers of phase-only metasurfaces, each of which contains 1024 units, mimicking a fully connected neural network through the diffraction of light fields. The classification task of handwriting digits from the ‘0’ to ‘5’ dataset is verified, with an accuracy of over 90% on the blind test dataset, as well as demonstrated by the full-wave simulation. Furthermore, the performance of the more complex animal image classification task is also validated by increasing the number of neurons to enhance the connectivity of the neural network. This study may provide a possible solution for practical applications such as biomedical detection, image processing, and machine vision based on all-optical computing. Full article

Analyzing point clouds with neural networks is a current research hotspot. In order to analyze the 3D geometric features of point clouds, most neural networks improve the network performance by adding local geometric operators and trainable parameters. However, deep learning usually requires a large amount of computational resources for training and inference, which poses challenges to hardware devices and energy consumption. Therefore, some researches have started to try to use a nonparametric approach to extract features. Point-NN combines nonparametric modules to build a nonparametric network for 3D point cloud analysis, and the nonparametric components include operations such as trigonometric embedding, farthest point sampling (FPS), k-nearest neighbor (k-NN), and pooling. However, Point-NN has some blindness in feature embedding using the trigonometric function during feature extraction. To eliminate this blindness as much as possible, we utilize a nonparametric energy function-based attention mechanism (ResSimAM). The embedded features are enhanced by calculating the energy of the features by the energy function, and then the ResSimAM is used to enhance the weights of the embedded features by the energy to enhance the features without adding any parameters to the original network; Point-NN needs to compute the similarity between each feature at the naive feature similarity matching stage; however, the magnitude difference of the features in vector space during the feature extraction stage may affect the final matching result. We use the Squash operation to squeeze the features. This nonlinear operation can make the features squeeze to a certain range without changing the original direction in the vector space, thus eliminating the effect of feature magnitude, and we can ultimately better complete the naive feature matching in the vector space. We inserted these modules into the network and build a nonparametric network, Point-Sim, which performs well in 3D classification tasks. Based on this, we extend the lightweight neural network Point-SimP by adding some trainable parameters for the point cloud classification task, which requires only 0.8 M parameters for high performance analysis. Experimental results demonstrate the effectiveness of our proposed algorithm in the point cloud shape classification task. The corresponding results on ModelNet40 and ScanObjectNN are 83.9% and 66.3% for 0 M parameters—without any training—and 93.3% and 86.6% for 0.8 M parameters. The Point-SimP reaches a test speed of 962 samples per second on the ModelNet40 dataset. The experimental results show that our proposed method effectively improves the performance on point cloud classification networks. Full article

In personalized nutrition, specific recommendations are often based on extensive phenotyping. In the world of microbiome research, classification is often based on the bacteriological composition of gut microbiota and enterotypes. We investigated if there is a possibility of translating outcomes from an intestinal screening platform to an intervention study that makes use of phenotyping. A 12-week double-blind, randomized, placebo-controlled, crossover intervention study (8-week wash-out period) with a dietary fiber mixture of acacia gum and carrot powder (ratio 3.33:1) was performed in healthy volunteers (N = 54, 45–70 years, BMI 27.3 ± 1.4) to modulate their microbiome. Fecal samples were collected every 4 weeks during the 32-week study period. Before and after the intervention a standardized mixed meal challenge was performed and plasma samples were taken (0, 30, 60, 120, and 240 min). Postprandial responses were used for sub-group cluster analysis to identify the metabolic phenotype. The individual participants’ samples were cultured anaerobically for 24 h with the mixture and the individual fibers. Compositional 16s rRNA data of exposed in vitro (24 h) and in vivo samples (4, 8, and 12 weeks) was compared and linked to the metabolic cluster analysis. The comparison between the clinical intervention’s effect on microbiota composition after 12 weeks and a single 24 h exposure in vitro showed a statistically significant association in microbiome effects between in vivo and in vitro exposures (p < 0.05) for the fiber intervention. Analysis of the metabolic postprandial responses revealed a division between improvement and deterioration in response to the fiber intervention indicating two distinct clusters (metabolic phenotypes). Cluster 1 contained the lowest triglycerides-, total cholesterol-, and non-esterified fatty acids responses, while cluster 2 contained the highest triglycerides- and total cholesterol responses. Interestingly, the beta diversity of the microbiota was linked to these two clusters, resembling two different responses to the fiber intervention. Our study in healthy individuals demonstrates that a short-term in vitro exposure of individual microbiome samples to the fiber mixture is predictive of a long-term in vivo effect and relates to a distinct phenotypic cluster. This paves the way for using the in vitro platform as a pre-screen for intervention studies. Full article

Hypertensive retinopathy (HR) results from the microvascular retinal changes triggered by hypertension, which is the most common leading cause of preventable blindness worldwide. Therefore, it is necessary to develop an automated system for HR detection and evaluation using retinal images. We aimed to propose an automated approach to identify and categorize the various degrees of HR severity. A new network called the spatial convolution module (SCM) combines cross-channel and spatial information, and the convolution operations extract helpful features. The present model is evaluated using publicly accessible datasets ODIR, INSPIREVR, and VICAVR. We applied the augmentation to artificially increase the dataset of 1200 fundus images. The different HR severity levels of normal, mild, moderate, severe, and malignant are finally classified with the reduced time when compared to the existing models because in the proposed model, convolutional layers run only once on the input fundus images, which leads to a speedup and reduces the processing time in detecting the abnormalities in the vascular structure. According to the findings, the improved SVM had the highest detection and classification accuracy rate in the vessel classification with an accuracy of 98.99% and completed the task in 160.4 s. The ten-fold classification achieved the highest accuracy of 98.99%, i.e., 0.27 higher than the five-fold classification accuracy and the improved KNN classifier achieved an accuracy of 98.72%. When computation efficiency is a priority, the proposed model’s ability to quickly recognize different HR severity levels is significant. Full article

Diabetic Retinopathy (DR) is one of the most common microvascular complications of diabetes. Diabetic Macular Edema (DME) is a concomitant symptom of DR. As the grade of lesion of DR and DME increase, the possibility of blindness can also increase significantly. In order to take the early interventions as soon as possible to reduce the likelihood of blindness, it is necessary to perform both DR and DME grading. We design a joint grading model based on multi-task learning and multi-branch networks (MaMNet) for DR and DME grading. The model mainly includes a multi-branch network (MbN), a feature fusion module, and a disease classification module. The MbN is formed by four branch structures, which can extract the low-level feature information of DME and DR in a targeted way; the feature fusion module is composed of a self-feature extraction module (SFEN), cross-feature extraction module (CFEN) and atrous spatial pyramid pooling module (ASPP). By combining various features collected from the aforementioned modules, the feature fusion module can provide more thorough discriminative features, which benefits the joint grading accuracy. The ISBI-2018-IDRiD challenge dataset is used to evaluate the performance of the proposed model. The experimental results show that based on the multi-task strategy the two grading tasks of DR and DME can provide each other with additional useful information. The joint accuracy of the model, the accuracy of DR and the accuracy of DME are 61.2%, 64.1% and 79.4% respectively. Full article

Diabetic retinopathy (DR), a consequence of diabetes, is one of the prominent contributors to blindness. Effective intervention necessitates accurate classification of DR; this is a need that computer vision-based technologies address. However, using large-scale deep learning models for DR classification presents difficulties, especially when integrating them into devices with limited resources, particularly in places with poor technological infrastructure. In order to address this, our research presents a knowledge distillation-based approach, where we train a fusion model, composed of ResNet152V2 and Swin Transformer, as the teacher model. The knowledge learned from the heavy teacher model is transferred to the lightweight student model of 102 megabytes, which consists of Xception with a customized convolutional block attention module (CBAM). The system also integrates a four-stage image enhancement technique to improve the image quality. We compared the model against eight state-of-the-art classifiers on five evaluation metrics; the experiments show superior performance of the model over other methods on two datasets (APTOS and IDRiD). The model performed exceptionally well on the APTOS dataset, achieving 100% accuracy in binary classification and 99.04% accuracy in multi-class classification. On the IDRiD dataset, the results were 98.05% for binary classification accuracy and 94.17% for multi-class accuracy. The proposed approach shows promise for practical applications, enabling accessible DR assessment even in technologically underdeveloped environments. Full article

Blind modulation classification (MC) is an integral part of designing an adaptive or intelligent transceiver for future wireless communications. Blind MC has several applications in the adaptive and automated systems of sixth generation (6G) communications to improve spectral efficiency and power efficiency, and reduce latency. It will become a integral part of intelligent software-defined radios (SDR) for future communication. In this paper, we provide various MC techniques for orthogonal frequency division multiplexing (OFDM) signals in a systematic way. We focus on the most widely used statistical and machine learning (ML) models and emphasize their advantages and limitations. The statistical-based blind MC includes likelihood-based (LB), maximum a posteriori (MAP) and feature-based methods (FB). The ML-based automated MC includes k-nearest neighbors (KNN), support vector machine (SVM), decision trees (DTs), convolutional neural networks (CNNs), recurrent neural networks (RNNs), and long short-term memory (LSTM) based MC methods. This survey will help the reader to understand the main characteristics of each technique, their advantages and disadvantages. We have also simulated some primary methods, i.e., statistical- and ML-based algorithms, under various constraints, which allows a fair comparison among different methodologies. The overall system performance in terms bit error rate (BER) in the presence of MC is also provided. We also provide a survey of some practical experiment works carried out through National Instrument hardware over an indoor propagation environment. In the end, open problems and possible directions for blind MC research are briefly discussed. Full article