Search Results (45)

Objective: To develop and evaluate artificial intelligence (AI)-based models for detecting laryngeal cancer using laryngoscope images. Methods: Two deep learning models were designed. The first identified and selected vocal cord images from laryngoscope datasets; the second localized laryngeal cancer within the selected images. Both employed FCN–ResNet101. Datasets were annotated by otolaryngologists, preprocessed (cropping, normalization), and augmented (horizontal/vertical flip, grid distortion, color jitter). Performance was assessed using Intersection over Union (IoU), Dice score, accuracy, precision, recall, F1 score, and per-image inference time. Results: The vocal cord selection model achieved a mean IoU of 0.6534 and mean Dice score of 0.7692, with image-level accuracy of 0.9972. The laryngeal cancer model achieved a mean IoU of 0.6469 and mean Dice score of 0.7515, with accuracy of 0.9860. Real-time inference was observed (0.0244–0.0284 s/image). Conclusions: By integrating a vocal cord selection model with a lesion detection model, the proposed platform enables accurate and fast detection of laryngeal cancer from laryngoscope images under the current experimental setting. Full article

Structural deterioration inevitably leads to defects in buildings. It is primarily caused by environmental exposure, material ageing, and long-term service conditions, whereas defects such as poor soil compaction arise from improper construction practices rather than deterioration mechanisms. Major concrete defects include missing portions such as cracking, corrosion, dents, blemishes, and spalling. Failure to identify minor issues can lead to serious problems, which become more expensive and difficult to repair, as well as poorer overall building performance. Traditional structural assessment methods, such as visual inspections and non-destructive testing are typically used for periodic condition evaluation, whereas SHM involves continuous or long-term monitoring using sensor-based systems. However, such approaches can be manual, costly, dangerous, and biased. In order to overcome these limitations, contemporary SHM systems combine traditional approaches with building information modelling (BIM) and artificial intelligence (AI). Different AI algorithms are used, including SVM, random forest, regression, and KNN for machine learning and decision trees; random forest, K-means clustering, CNN, U-Net, ResNet, FCN, VGG16, and DeepLabv3+ for deep learning. This review will survey both the traditional and novel approaches in the field of SHM and the recent advancements. Full article

Earth Observation increasingly uses machine learning to evaluate and monitor the environment. However, the potential of deep learning for studying wilderness is an under-explored frontier. This study aims to give insights into using different architectures (ResNet18, ResNet50, U-Net, DeepLabV3, and FCN), batch sizes (small, medium, and large), and spectral setups (RGB, RGB+NIR, full spectrum) for the classification and semantic segmentation of Sentinel-2 images. The focus is on optimising performance over accuracy using limited computational resources and pre-trained networks widely from the AI community. Experiments are performed on the AnthroProtect dataset, which was developed explicitly for this purpose. Results show that when computation resources are a concern, ResNet18 with 64 or 256 batch size is an optimal configuration for image classification. The U-Net is a sub-optimal solution for semantic segmentation, but our experiments did not identify a clear optimality for the batch size. Finally, different spectral setups highlight no significant impact on the data processing, thus raising critical thinking on the usefulness of neural networks in Earth Observation that are pre-trained with generic data like ImageNet, which is widely used in the AI community. Full article

Tailings storage facilities (TSFs) have experienced numerous global failures, many linked to active deposition on tailings beaches. Understanding these processes is vital for effective management. As deposition alters surface elevation, developing an explainable model to predict the changes can enhance insight into deposition dynamics and support proactive TSF management. This study applies deep learning (DL) to predict surface elevation changes in tailings storage facilities (TSFs) from high-resolution digital elevation models (DEMs) generated from UAV photogrammetry. Three DL architectures, including multilayer perceptron (MLP), fully convolutional network (FCN), and residual network (ResNet), were evaluated across spatial patch sizes of 64 × 64, 128 × 128, and 256 × 256 pixels. The results show that incorporating broader spatial contexts improves predictive accuracy, with ResNet achieving an $R2$ of 0.886 at the 256 × 256 scale, explaining nearly 89% of the variance in observed deposition patterns. To enhance interpretability, SHapley Additive exPlanations (SHAP) were applied, revealing that spatial coordinates and curvature exert the strongest influence, linking deposition patterns to discharge distance and microtopographic variability. By prioritizing predictive performance while providing mechanistic insight, this framework offers a practical and quantitative tool for reliable TSF monitoring and management. Full article

Pulmonary embolism (PE) on CT pulmonary angiography (CTPA) demands rapid, accurate assessment, yet small, low-contrast clots in distal arteries remain challenging. We benchmarked ten fully convolutional network (FCN) backbones and introduced Consensus Intersection-Optimized Fusion (CIOF)—a K-of-M, pixel-wise mask fusion with the voting threshold K* selected on training patients to maximize IoU. Using the FUMPE cohort (35 patients; 12,034 slices) with patient-based random splits (18 train, 17 test), we trained five FCN architectures (each with Adam and SGDM) and evaluated segmentation with IoU, Dice, FNR/FPR, and latency. CIOF achieved the best overall performance (mean IoU 0.569; mean Dice 0.691; FNR 0.262), albeit with a higher runtime (~63.7 s per case) because all ten models are executed and fused; the strongest single backbone was Inception-ResNetV2 + SGDM (IoU 0.530; Dice 0.648). Stratified by embolization ratio, CIOF remained superior across <10⁻⁴, 10⁻⁴–10⁻³, and >10⁻³ clot burdens, with mean IoU/Dice = 0.238/0.328, 0.566/0.698, and 0.739/0.846, respectively—demonstrating gains for tiny, subsegmental emboli. These results position CIOF as an accuracy-oriented, interpretable ensemble for offline or second-reader use, while faster single backbones remain candidates for time-critical triage. Full article

Digital epidermal microscopic (DEM) images offer the potential to quantitatively analyze skin aging at the microscopic level. However, stochastic complexity, local highlights, and low contrast in DEM images pose significant challenges to accurate segmentation. This study evaluated eight deep learning models to identify methods capable of accurately segmenting complex DEM images while meeting diverse performance requirements. To this end, this study first constructed a manually labeled DEM image dataset. Then, eight deep learning models (FCN-8s, SegNet, UNet, ResUNet, NestedUNet, DeepLabV3+, TransUNet, and AttentionUNet) were systematically evaluated for their performance in DEM image segmentation. Our experimental findings show that AttentionUNet achieves the highest segmentation accuracy, with a DSC of 0.8696 and an IoU of 0.7703. In contrast, FCN-8s is a better choice for efficient segmentation due to its lower parameter count (18.64 M) and efficient inference speed (GPU time 37.36 ms). FCN-8s and NestedUNet show a better balance between accuracy and efficiency when assessed across metrics like segmentation accuracy, model size, and inference time. Through a systematic comparison of eight deep learning models, this study identifies superior methods for segmenting skin furrows and ridges in DEM images. This work lays the foundation for subsequent applications, such as analyzing skin aging through furrow and ridge features. Full article

Machine Learning models, more specifically Artificial Neural Networks, are transforming medical imaging by enabling precise liver segmentation, a crucial task for diagnosing and treating liver diseases. However, these models often face challenges in adapting to diverse clinical data sources as differences in dataset volume, resolution, and origin impact generalization and performance. This study introduces a Private Data Incrementalization, a data-centric approach to enhance the adaptability of Artificial Neural Networks by progressively exposing them to varied clinical data. As the target of this study is not to propose a new image segmentation model, the existing medical imaging segmentation models—including U-Net, ResUNet++, Fully Convolutional Network, and a modified algorithm based on the Conditional Bernoulli Diffusion Model—are used. The study evaluates these four models using a curated private dataset of computed tomography scans from Coimbra University Hospital, supplemented by two public datasets, 3D-IRCADb01 and CHAOS. The Private Data Incrementalization method systematically increases the volume and diversity of training data, simulating real-world conditions where models must handle varied imaging contexts. Pre-processing and post-processing stages, incremental training, and performance evaluations reveal that structured exposure to diverse datasets improves segmentation performance, with ResUNet++ achieving the highest accuracy (0.9972) and Dice Similarity Coefficient (0.9449), and the best Average Symmetric Surface Distance (0.0053 mm), demonstrating the importance of dataset diversity and volume for segmentation models’ robustness and generalization. Private Data Incrementalization thus offers a scalable strategy for building resilient segmentation models, ultimately benefiting clinical workflows, patient care, and healthcare resource management by addressing the variability inherent in clinical imaging data. Full article

Augmented reality applications involving human interaction with virtual objects often rely on segmentation-based hand detection techniques. Semantic segmentation can then be enhanced with instance-specific information to model complex interactions between objects, but extracting such information typically increases the computational load significantly. This study proposes a training strategy that enables conventional semantic segmentation networks to preserve some instance information during inference. This is accomplished by introducing pixel weight maps into the loss calculation, increasing the importance of boundary pixels between instances. We compare two common fully convolutional network (FCN) architectures, U-Net and ResNet, and fine-tune the fittest to improve segmentation results. Although the resulting model does not reach state-of-the-art segmentation performance on the EgoHands dataset, it preserves some instance information with no computational overhead. As expected, degraded segmentations are a necessary trade-off to preserve boundaries when instances are close together. This strategy allows approximating instance segmentation in real-time using non-specialized hardware, obtaining a unique blob for an instance with an intersection over union greater than 50% in 79% of the instances in our test set. A simple FCN, typically used for semantic segmentation, has shown promising instance segmentation results by introducing per-pixel weight maps during training for light-weight applications. Full article

The purpose of this research is to contribute to the development of approaches for the classification and segmentation of various gastrointestinal (GI) cancer diseases, such as dyed lifted polyps, dyed resection margins, esophagitis, normal cecum, normal pylorus, normal Z line, polyps, and ulcerative colitis. This research is relevant and essential because of the current challenges related to the absence of efficient diagnostic tools for early diagnostics of GI cancers, which are fundamental for improving the diagnosis of these common diseases. To address the above challenges, we propose a new hybrid segmentation model, U-MaskNet, which is a combination of U-Net and Mask R-CNN models. Here, U-Net is utilized for pixel-wise classification and Mask R-CNN for instance segmentation, together forming a solution for classifying and segmenting GI cancer. The Kvasir dataset, which includes 8000 endoscopic images of various GI cancers, is utilized to validate the proposed methodology. The experimental results clearly demonstrated that the novel proposed model provided superior segmentation compared to other well-known models, such as DeepLabv3+, FCN, and DeepMask, as well as improved classification performance compared to state-of-the-art (SOTA) models, including LeNet-5, AlexNet, VGG-16, ResNet-50, and the Inception Network. The quantitative analysis revealed that our proposed model outperformed the other models, achieving a precision of 98.85%, recall of 98.49%, and F1 score of 98.68%. Additionally, the novel model achieved a Dice coefficient of 94.35% and IoU of 89.31%. Consequently, the developed model increased the accuracy and reliability in detecting and segmenting GI cancer, and it was proven that the proposed model can potentially be used for improving the diagnostic process and, consequently, patient care in the clinical environment. This work highlights the benefits of integrating the U-Net and Mask R-CNN models, opening the way for further research in medical image segmentation. Full article

To address the challenges that convolutional neural networks (CNNs) face in extracting small objects and handling class imbalance in remote sensing imagery, this paper proposes a novel spatial contextual information and multiscale feature fusion encoding–decoding network, SCIMF-Net. Firstly, SCIMF-Net employs an improved ResNeXt-101 deep backbone network, significantly enhancing the extraction capability of small object features. Next, a novel PMFF module is designed to effectively promote the fusion of features at different scales, deepening the model’s understanding of global and local spatial contextual information. Finally, introducing a weighted joint loss function improves the SCIMF-Net model’s performance in extracting LULC information under class imbalance conditions. Experimental results show that compared to other CNNs such as Res-FCN, U-Net, SE-U-Net, and U-Net++, SCIMF-Net improves PA by 0.68%, 0.54%, 1.61%, and 3.39%, respectively; MPA by 2.96%, 4.51%, 2.37%, and 3.45%, respectively; and MIOU by 3.27%, 4.89%, 4.2%, and 5.68%, respectively. Detailed comparisons of locally visualized LULC information extraction results indicate that SCIMF-Net can accurately extract information from imbalanced classes and small objects. Full article

Accurately mapping the surface water bodies through remote sensing technology is of great significance for water resources management, flood monitoring, and drought monitoring. At present, many scholars at home and abroad carry out research on deep learning image recognition algorithms based on convolutional neural networks, and a variety of variant-based convolutional neural networks are proposed to be applied to extract water bodies from remote sensing images. However, due to the low depth of convolutional layers employed and underutilization of water spectral feature information, most of the water body extraction methods based on convolutional neural networks (CNNs) for remote sensing images are limited in accuracy. In this study, we propose a novel surface water automatic extraction method based on the convolutional neural network (CRAUnet++) for Sentinel-2 images. The proposed method includes three parts: (1) substituting the feature extractor of the original Unet++ with ResNet34 to enhance the network’s complexity by increasing its depth; (2) Embedding the Spatial and Channel ‘Squeeze and Excitation’ (SCSE) module into the up-sampling stage of the network to suppress background features and amplify water body features; (3) adding the vegetation red edge-based water index (RWI) into the input data to maximize the utilization of water body spectral information of Sentinel-2 images without increasing the data processing time. To verify the performance and accuracy of the proposed algorithm, the ablation experiment under four different strategies and comparison experiment with different algorithms of RWI, FCN, SegNet, Unet, and DeepLab v3+ were conducted on Sentinel-2 images of the Poyang Lake. The experimental result shows that the precision, recall, F1, and IoU of CRAUnet++ are 95.99%, 96.41%, 96.19%, and 92.67%, respectively. CRAUnet++ has a good performance in extracting various types of water bodies and suppressing noises because it introduces SCSE attention mechanisms and combines surface water spectral features from RWI, exceeding that of the other five algorithms. The result demonstrates that CRAUnet++ has high validity and reliability in extracting surface water bodies based on Sentinel-2 images. Full article

With the development of satellite remote sensing technology, a substantial quantity of remote sensing data can be obtained every day, but the ability to extract information from these data remains poor, especially regarding intelligent extraction models for vegetation information in mountainous areas. Because the features of remote sensing images (such as spectral, textural and geometric features) change with changes in illumination, viewing angle, scale and spectrum, it is difficult for a remote sensing intelligent interpretation model with a single data source as input to meet the requirements of engineering or large-scale vegetation information extraction and updating. The effective use multi-source, multi-resolution and multi-type data for remote sensing classification is still a challenge. The objective of this study is to develop a highly intelligent and generalizable classification model of mountain vegetation utilizing multi-source remote sensing data to achieve accurate vegetation extraction. Therefore, a multi-channel semantic segmentation model based on deep learning, FCN-ResNet, is proposed to integrate the features and textures of multi-source, multi-resolution and multi-temporal remote sensing data, thereby enhancing the differentiation of different mountain vegetation types by capturing their characteristics and dynamic changes. In addition, several sets of ablation experiments are designed to investigate the effectiveness of the model. The method is validated on Mt. Taibai (part of the Qinling-Daba Mountains), and the pixel accuracy (PA) of vegetation classification reaches 85.8%. The results show that the proposed multi-channel semantic segmentation model can effectively discriminate different vegetation types and has good intelligence and generalization ability in different mountainous areas with similar vegetation distributions. The multi-channel semantic segmentation model can be used for the rapid updating of vegetation type maps in mountainous areas. Full article

Lightning Electromagnetic Pulses, or LEMPs, propagate in the Earth–ionosphere waveguide and can be detected remotely by ground-based lightning electric field sensors. LEMPs produced by different types of lightning processes have different signatures. A single thunderstorm can produce thousands of LEMPs, which makes their classification virtually impossible to carry out manually. The lightning classification is important to distinguish the types of thunderstorms and to know their severity. Lightning type is also related to aerosol concentration and can reveal wildfires. Artificial Intelligence (AI) is a good approach to recognizing patterns and dealing with huge datasets. AI is the general denomination for different Machine Learning Algorithms (MLAs) including deep learning and others. The constant improvements in the AI field show us that most of the Lightning Location Systems (LLS) will soon incorporate those techniques to improve their performance in the lightning-type classification task. In this study, we assess the performance of different MLAs, including a SVM (Support Vector Machine), MLP (Multi-Layer Perceptron), FCN (Fully Convolutional Network), and Residual Neural Network (ResNet) in the task of LEMP classification. We also address different aspects of the dataset that can interfere with the classification problem, including data balance, noise level, and LEMP recorded length. Full article

This paper proposes a multimodal fusion 3D target detection algorithm based on the attention mechanism to improve the performance of 3D target detection. The algorithm utilizes point cloud data and information from the camera. For image feature extraction, the ResNet50 + FPN architecture extracts features at four levels. Point cloud feature extraction employs the voxel method and FCN to extract point and voxel features. The fusion of image and point cloud features is achieved through regional point fusion and voxel fusion methods. After information fusion, the Coordinate and SimAM attention mechanisms extract fusion features at a deep level. The algorithm’s performance is evaluated using the DAIR-V2X dataset. The results show that compared to the Part-A2 algorithm; the proposed algorithm improves the mAP value by 7.9% in the BEV view and 7.8% in the 3D view at IOU = 0.5 (cars) and IOU = 0.25 (pedestrians and cyclists). At IOU = 0.7 (cars) and IOU = 0.5 (pedestrians and cyclists), the mAP value of the SECOND algorithm is improved by 5.4% in the BEV view and 4.3% in the 3D view, compared to other comparison algorithms. Full article

Deep learning networks might require re-training for different datasets, consuming significant manual labeling and training time. Transfer learning uses little new data and training time to enable pre-trained network segmentation in relevant scenarios (e.g., different vegetation images in rainy and dry seasons); however, existing transfer learning methods lack systematicity and controllability. So, an MTPI method (Maximum Transfer Potential Index method) was proposed to find the optimal conditions in data and feature quantity for transfer learning (MTPI conditions) in this study. The four pre-trained deep networks (Seg-Net (Semantic Segmentation Networks), FCN (Fully Convolutional Networks), Mobile net v2, and Res-Net 50 (Residual Network)) using the rainy season dataset showed that Res-Net 50 had the best accuracy with 93.58% and an WIoU (weight Intersection over Union) of 88.14%, most worthy to transfer training in vegetation segmentation. By obtaining each layer’s TPI performance (Transfer Potential Index) of the pre-trained Res-Net 50, the MTPI method results show that the 1000-TDS and 37-TP were estimated as the best training speed with the smallest dataset and a small error risk. The MTPI transfer learning results show 91.56% accuracy and 84.86% WIoU with 90% new dataset reduction and 90% iteration reduction, which is informative for deep networks in segmentation tasks between complex vegetation scenes. Full article