Search Results (4,553)

Reliable defect detection in motor production lines is important for maintaining manufacturing quality. This study investigates the feasibility of a deep learning-based airborne acoustic screening approach for controlled no-load end-of-line induction motor inspection. Acoustic signals collected under controlled no-load test conditions were used to evaluate Feedforward Neural Networks (FNNs), Convolutional Neural Networks (CNNs), Recurrent Neural Networks (RNNs), and pre-trained models including ResNet50, MobileNetV2, and EfficientNetB0 for detecting rotor unbalance, assembly-induced bearing abnormalities, and their combination. Experimental results show that several models achieved up to 96% classification accuracy, depending on the selected architecture and feature representation. In cross-motor external validation using an unseen motor from the same manufacturer, the CNN with MFCC features achieved the best performance with 96% accuracy and 97% precision, recall, and F1-score, while ResNet50 with spectrogram inputs achieved 93% accuracy and 92% F1-score. These results demonstrate that airborne motor acoustic signals contain discriminative defect-related information under controlled no-load conditions and support the feasibility of low-cost, non-contact airborne acoustic sensing as a complementary screening approach for rapid end-of-line motor inspection. Full article

One of the most challenging tasks in pediatric medicine is bone age estimation from hand radiographs. Traditional bone age estimation approaches show limited performance across different demographic groups, making diagnostics prone to misclassification of growth abnormalities. On the other hand, convolutional neural networks have demonstrated remarkable performance for many computer vision tasks, including bone age estimation. These models automatically extract and learn meaningful patterns capturing structural variations in bones. However, these models still present limited generalization capabilities, especially across different demographic populations, and are rarely evaluated beyond the public datasets on which they are trained. This paper presents a fusion architecture, F-DenseNet121, together with a systematic evaluation of its generalization across demographic populations. The employed methodology incorporates an anatomical segmentation strategy inspired by the Tanner–Whitehouse 3 (TW3) framework, combined with feature learning from the RSNA Pediatric Bone Age Challenge dataset. Instead of training a standalone convolutional model, the proposed methodology considers a convolutional network for each anatomical segment. This mechanism allows the model to learn localized skeletal patterns, reducing the influence of irrelevant structures. During training and validation with the RSNA dataset, the proposed F-DenseNet121 obtained a Mean Absolute Error (MAE) of 5.77 months during internal validation, a figure comparable to several reported convolutional models under their respective internal validation protocols. F-DenseNet121 was also evaluated using an independent 10.8% RSNA test validation subset, obtaining an MAE of 13.70 months, a result that remains substantially higher than published state-of-the-art benchmarks (4.2–6.2 months) and reveals a substantial generalization gap between internal validation and independent testing. To further examine this gap, external validation was performed using radiographs from Mexican patients. In this test, all convolutional models showed a significant performance difference between the public RSNA dataset and the clinical data from Mexican patients, with F-DenseNet121 and F-InceptionV3 achieving statistically indistinguishable external performance among the evaluated backbones. Rather than positioning these results as evidence of state-of-the-art accuracy, this study highlights that internal validation performance can substantially overestimate real-world reliability, and underscores the value of rigorous, multi-architecture comparison and external clinical validation for assessing the true applicability of automated bone age estimation systems. Full article

Agile project monitoring commonly relies on numerical metrics and visual artifacts, such as Burndown charts, to assess Sprint progress and identify potential deviations. However, most automated approaches focus on structured data, while the visual patterns contained in agile charts remain underexplored. This paper presents an exploratory study on visual-based Sprint evaluation using convolutional neural networks and agile project metrics. The proposed approach uses Burndown and Completed vs. Uncompleted Work chart (TTvsNT) images to classify Sprint performance into four categories: Poor, Regular, Good, and Excellent. A transfer-learning strategy based on MobileNetV2 was applied, including image preprocessing, Sprint-level data partitioning, two-phase training, and multiclass evaluation. The model achieved an overall accuracy of 70.33% on the evaluation set. Class-level results showed better performance for the Poor and Excellent categories, while the intermediate classes presented greater ambiguity. The main contribution of this study lies in evaluating Sprint monitoring charts as a complementary visual representation to traditional metric-based models. The findings provide preliminary evidence that these images contain useful performance-related patterns; however, the limited dataset size and current accuracy do not support production-level deployment. Further research with larger datasets, additional architectures, and multimodal approaches is required. Full article

Background: Breast cancer brain metastases (BCBMs) exhibit notable receptor discordance between primary tumors and metastases, yet obtaining intracranial biopsies is rarely practical. We created an interpretable deep learning radiogenomic model to predict estrogen receptor (ER), progesterone receptor (PR), and human epidermal growth factor receptor 2 (HER2) status from MRI scans in BCBM cases. Methods: A total of 241 post-contrast T1-weighted MRIs from 142 patients were analyzed. We developed a mask-free 3D Residual Neural Network (ResNet) ensemble trained directly on cropped, bias-corrected brain images, with comprehensive 3D geometric and intensity augmentations. The model was optimized in a multi-label setting using Asymmetric Focal Loss. Hyperparameters were fine-tuned via Bayesian optimization, and class probabilities were calibrated. Additionally, Integrated Gradients (IG) offered voxel-level saliency maps. Results: Our ResNet ensemble model achieved a micro-averaged AUROC of 0.74 and a macro-averaged AUROC of 0.67. Receptor-specific AUROCs were 0.57 for ER, 0.79 for PR, and 0.64 for HER2. The F1-scores were 0.56, 0.62, and 0.88 for ER, PR, and HER2, respectively. The held-out cohort contained only four HER2-negative patients, so threshold-dependent HER2 metrics are strongly prevalence-driven and AUROC is the more appropriate summary. Qualitatively inspected saliency maps were concentrated on enhancing metastases with limited background attribution. Conclusions: This proof-of-concept study shows that a segmentation-free, interpretable 3D Convolutional Neural Network (CNN) can be trained to capture receptor-associated patterns in post-contrast T1-weighted MRI of BCBMs. Accuracy was modest relative to published radiomics models, and the mask-free, single-sequence design is offered as a methodological contribution rather than a performance gain. These findings are preliminary, do not establish clinical utility, and require validation in larger, multi-center cohorts. Full article

On edge devices, convolutional neural network (CNN) inference is bottlenecked mainly by memory bandwidth, owing to the frequent memory accesses to feature maps and parameters. To address this challenge, we propose a memory-efficient hardware accelerator for depthwise separable convolution that minimizes off-chip memory traffic and parameter storage. The proposed architecture employs three key techniques: (1) a 64-bit run-length coding (RLC) packet compression that exploits feature-map sparsity after ReLU, (2) a mixed-precision scheme that represents feature maps and weights at different precisions, and (3) separated depthwise and pointwise convolution units. In particular, feature maps are transferred in RLC-compressed form, which reduces the amount of data exchanged with the host. The compressed data are decoded row by row, so the on-chip buffers hold only the rows required for computation rather than a complete feature map. In software simulation on ImageNet, the mixed-precision scheme reduced the parameter storage by 49.22% at a cost of 6.24 percentage points (pp) in Top-1 accuracy, and the RLC reduced the data by up to 56.07% in the deeper layers. Implemented on a Xilinx ZCU-104 FPGA, the proposed accelerator performs the depthwise separable convolution with a small number of logic resources and on-chip memory, confirming its feasibility for resource-constrained edge devices. Full article

Quality control in green coffee bean production is critical for food safety and economic sustainability. A persistent gap in existing automated inspection systems is their inability to capture both sides of each bean, which leads to systematic under-detection of surface defects. This work presents three contributions: (i) a novel mechatronic apparatus that mechanically guarantees dual-sided imaging of every bean, (ii) a public 12-class dataset of green coffee bean defects, and (iii) an embedded, real-time inspection pipeline validated on low-cost hardware. The apparatus sequentially presents each bean, from a standard 350 g sample, to two 16-megapixel cameras under controlled LED illumination. A dataset of 9600 images spanning 12 classes (11 defects and 1 normal) was generated from expert-classified samples and enriched through data augmentation. Four convolutional neural network (CNN) architectures, VGG-16, VGG-19, ResNet-50 and YOLOv8, were trained and benchmarked using precision, recall, F1-score and mean average precision. YOLOv8 achieved the best overall performance, with a precision of 97.4%, a recall of 99.6%, an F1-score of 0.930 and a mean average precision of 96.5%, outperforming VGG-16 (accuracy 86.07%), VGG-19 (accuracy 67.03%) and ResNet-50 (accuracy 87.76%). Dual-sided acquisition raised mean per-class detection accuracy from 0.727 to 0.908, a relative gain of 25.7% over an equivalent single-sided configuration. Deployed in real-time “track” mode on a Raspberry Pi 4, the system simultaneously classifies defects and counts beans by category, processing a 350 g sample in approximately 38 min. Combining mechanical innovation with lightweight deep learning enables practical, scalable, and cost-effective quality control for laboratories specialized in coffee analysis. Full article

Background/Objectives: Brain tumor diagnosis using magnetic resonance imaging (MRI) plays an important role in clinical decision-making and treatment planning. However, manual interpretation of MRI scans is time-consuming and may result in variations among radiologists. This study aims to develop an automated multiclass brain tumor classification framework by integrating deep learning-based feature extraction with the Bitterling Fish Optimization (BFO) algorithm for effective feature selection. Methods: MRI images were first subjected to preprocessing to prepare them for deep learning analysis. Several pretrained convolutional neural network (CNN) architectures, including VGG16, VGG19, InceptionV3, ResNet50, EfficientNet, MobileNet, and ShuffleNet, were employed to extract informative deep features from the MRI images. The extracted features were then optimized using the BFO algorithm, which selected the most relevant features while reducing feature redundancy. The selected features were subsequently used for multiclass brain tumor classification. Model performance was evaluated using sensitivity, specificity, precision, accuracy, F1-score, and area under the curve (AUC). Results: The experimental results demonstrated that BFO-based feature selection improved the classification performance of the evaluated CNN architectures compared with their corresponding models without feature selection. Among the investigated CNN–optimizer combinations, the BFO-ShuffleNet framework achieved the best overall performance, obtaining 98.98% sensitivity, 98.99% specificity, 98.99% precision, 99.00% accuracy, and a 98.99% F1-score. These results indicate that BFO effectively identified the most discriminative features and enhanced the classification capability of the ShuffleNet architecture. Conclusions: The proposed deep learning and BFO-based framework provide an accurate and efficient approach for automated multiclass brain tumor classification from MRI images. The findings demonstrate that combining pretrained CNN models with BFO-based feature selection can reduce feature redundancy and improve diagnostic performance. In particular, BFO-ShuffleNet demonstrated the highest classification performance and shows considerable potential as a computer-aided diagnostic tool to support radiologists in brain tumor assessment and clinical decision-making. Full article

Convolutional neural networks (CNNs), as an important branch of artificial intelligence, have a wide range of applications in security-critical scenarios, and their hardware accelerators are also constantly evolving. However, hardware failures can cause parameter bit flipping, and carefully designed bit-flipping attacks (BFAs) can also severely degrade their classification accuracy. This paper proposes a hardware-friendly security-reinforcement framework, which utilizes the representation characteristics of floating-point numbers on memory units to propose modulation and demodulation reinforcement methods for weights. On the computing unit, a data-threshold constraint reinforcement method is proposed, and a bounded ReLU is introduced to mitigate the impact of BFAs by imposing an upper bound on the activation function. Experimental results demonstrate that when the BER reaches 10⁻², the accuracies of LeNet-5, LeNet-3D, and ResNet-18 increase from 13.17%, 12.01%, and 10% to 93.21%, 59.12%, and 62.37%, respectively. Full article

Background/Objectives: Tissue fold artifacts observed in hematoxylin-eosin (H&E)- stained preparations used in histology education can complicate the assessment of normal tissue architecture and affect students’ accurate interpretation of microscopic structures. This study aimed to automatically detect and localize tissue fold artifacts in histology teaching preparations using deep learning methods. Methods: A total of 2127 hematoxylin and eosin (H&E)-stained histological images of brain, kidney, liver, small intestine, and testis tissues obtained at 10× magnification were used in the study. The dataset consisted of 899 clean/artifact-free images and 1228 images containing tissue fold artifacts. Seven deep learning architectures, including convolutional neural networks (CNN)-based models and a vision transformer-based model, were evaluated for image-level classification: ResNet18, ResNet50, DenseNet121, EfficientNet-B0, EfficientNet-B3, ConvNeXt-Tiny, and Swin-Tiny. Classification performance was evaluated using an organ-based testing approach, a slide-level train–test split in which images from the same histological slide group were retained within a single subset, a general image-level 80/20 train–test split, and five-fold cross-validation. A DeepLabV3-ResNet50-based segmentation model was trained using QuPath-prepared masks to determine fold regions at the pixel level. Grad-CAM was used for qualitative visualization and quantitative comparison with manually annotated fold regions. Results: All classification models showed excellent performance overall. The Swin-Tiny model was the most accurate in terms of general image-level classification, scoring 99.06% for accuracy, 99.19% for F1-score, and 99.98% for AUC. In slide-level classification, EfficientNet-B3 showed the best performance in terms of accuracy (99.54%) and AUC (99.99%). No pairwise statistical difference was found among the models using Holm adjustment. In the five-fold cross-validation setting, ResNet50 achieved the best performance with the accuracy of 98.73 ± 0.54%. In the detailed small-intestine error analysis, classification accuracy was 67.67%, with high sensitivity (98.59%) but low specificity (24.34%), mainly because of false-positive predictions. In the segmentation evaluation, the final DeepLabV3-ResNet50 model trained with combined BCE + Dice loss resulted in Dice of 0.7630 ± 0.2425 and IoU of 0.6661 ± 0.2577 on the independent test set. False positive segmentations were rare in 899 artifact-free images (only 0.33% of images had a tissue fold region of at least 1%). The quantitative Grad-CAM analysis showed poor spatial agreement with the manually annotated tissue fold masks (Dice = 0.2423; IoU = 0.1441). In the independent MPP10 dataset, ResNet50 achieved 89.63% accuracy and 88.49% F1-score. Conclusions: The suggested method was able to detect tissue fold artifacts as well as localize them in the teaching images of histology. The strong performance recorded at the slide level validates the internal findings, but the limited performance in the case of small intestine images and on the external dataset reveals that tissue structure remains a crucial factor. Full article

Retinitis pigmentosa (RP) is a rare inherited retinal disorder in which fundus changes may be subtle and heterogeneous, limiting detection from color fundus images. This study evaluated multiple machine learning architectures for binary-RP versus healthy-control classification, and developed a reproducible pipeline for research-oriented screening support. Three publicly available fundus datasets were combined, including 248 RP images and 1045 healthy controls. An 80/20 train–test split was used, with targeted data augmentation applied only to RP images in the training set to address class imbalance. ConvNeXt-Tiny, ResNet101V2, EfficientNet-B0, a baseline classifier, and custom shallow convolutional neural networks were compared using accuracy, precision, recall, F1-score, confusion matrices, and ROC/precision–recall analyses. A compact ShallowCNN provided the best sensitivity–performance trade-off. On the fixed image-level test set, Adam with a learning rate of 0.0005 reached 96.51% accuracy, while SGD with a learning rate of 0.001 achieved 98% RP recall, minimizing false negatives. The trained models were exported to ONNX and integrated into a Windows inference tool. The proposed framework provides an open, reproducible benchmark for technical evaluation, although external validation is required before clinical use. Full article

Automatic classification of Ataulfo mango (Mangifera indica L.) ripeness is essential to ensure consistent quality, standardize post-harvest processes, and reduce the subjectivity of traditional visual inspection, which is unreliable and error-prone. This paper aims to develop a computationally efficient image classification system for Ataulfo mango ripeness by combining explicit color and texture feature extraction with a traditional machine learning model, thereby reducing the high computational costs typically associated with deep convolutional architectures. For this purpose, a dataset containing 10,400 images was created and divided into four maturity categories: green-ripe, partially ripe, firm-ripe, and soft-ripe. We select an optimal Multilayer Perceptron trained on compact 33-dimensional feature vectors and compare its performance with classical machine learning algorithms and pretrained deep neural networks, including MobileNetV2, MobileNetV3, and ResNet18. Our proposal achieves an accuracy of 0.8821, a macro-F1 score of 0.8784, and an AUC of 0.9751, which are better than those of classical classifiers and MobileNet-family models, while reducing computational cost by three orders of magnitude (GFLOPs). The ResNet18 model achieved a 3.56% relative improvement in macro-F1 score compared to our proposal, but its computational cost increased by four orders of magnitude in GFLOPS. In all evaluated architectures, the remaining classification errors occur between adjacent maturity stages and likely reflect the visual similarity inherent in the continuous ripening process. These findings demonstrate that manual feature engineering and model selection via hyperparameter tuning remain highly competitive and more sustainable for low-cost edge implementations in agriculture. Full article

Lung cancer arises from mutations in lung cells, disrupting their normal growth cycle and leading to uncontrolled cell division. These rapidly dividing cells lose function and fail to form healthy lung tissue. Several factors contribute to the difficulty of diagnosing and classifying lung nodules, including the high degree of morphological heterogeneity and overlapping characteristics between benign and malignant nodules. Recently, deep learning models have been used in computed tomography (CT)-based lung nodule diagnosis and have demonstrated diagnostic efficiency comparable to that of radiologists. This study introduces LungCNET, a high-performance multi-layer deep convolutional neural network trained on chest CT images to improve lung lesion classification efficiency and accuracy significantly. The data for the Lung Cancer convolutional neural network (LungCNET) were derived from the IQ-OTH/NCCD CT scan dataset (1097 images from 110 cases), split into training (767 images), validation (109) and a held-out test partition (221) that played no role in training or model selection. This dataset encompasses three diagnostic categories: benign, malignant, and normal lung tissues. LungCNET was evaluated against fine-tuned benchmark models that are both established and widely used, spanning architectures introduced between 2014 and 2024, including VGG16, ResNet50, InceptionV3, MobileNetV2, and YOLOv11. On the held-out test partition, LungCNET reached a macro-averaged F1-score of 95.19%, with VGG16 at 94.09% and InceptionV3 at 92.28%; these three models performed comparably, and the separation between them is small relative to the resolution of a test set of this size. LungCNET was, however, the only model to exceed 90% F1-score across all three diagnostic classes simultaneously, and recorded the highest F1-score on the benign class (91.0%), the smallest and most frequently misclassified category, where two of the six models failed entirely. These results support LungCNET as a candidate tool for lung cancer diagnosis, subject to validation on larger and independently sourced datasets. Full article

Scoliosis is a complex three-dimensional spinal deformity that requires early detection, accurate radiological assessment, individualized treatment planning, and long-term monitoring. In recent years, artificial intelligence (AI) has emerged as a promising tool across multiple stages of scoliosis management, including screening, automated imaging analysis, deformity classification, prediction of progression, surgical planning, postoperative outcome assessment, and patient education. This narrative review summarizes current and emerging applications of AI in scoliosis, with particular emphasis on studies published after 2023. Deep learning algorithms, including convolutional neural networks, U-Net-based architectures, transformer models, and generative approaches, have demonstrated high accuracy in automated Cobb angle measurement, vertebral segmentation, coronal and sagittal parameter assessment, and radiation-free screening using surface topography or smartphone-based photographs. Machine learning models have also shown potential in predicting curve progression, treatment response, risk of postoperative complications, and patient-reported outcomes by integrating radiological, clinical, biomechanical, and, increasingly, multimodal data. In parallel, large language models and generative AI tools are being investigated for patient education, communication support, readability improvement, and research hypothesis generation. Despite these advances, important limitations remain, including limited external validation, dataset heterogeneity, potential algorithmic bias, insufficient interpretability, and incomplete integration into clinical workflows. Moreover, while AI systems show strong performance in automated measurement and screening tasks, their role in complex therapeutic decision-making, such as Lenke classification, fusion-level selection, and autonomous surgical planning, remains experimental. Overall, AI has the potential to improve the precision, efficiency, and personalization of scoliosis care; however, prospective multicentre studies, transparent reporting, explainable model design, and regulatory validation are essential before widespread clinical implementation. Full article

Adder neural networks remove multiplication from convolution, yet their direct L1-distance datapath still requires subtraction, absolute-value generation, and wide accumulation. We address this cost by mapping the online L1 operation to minimum selection and time-domain accumulation. The proposed accelerator processes a $3\times 3\times 16$ window for 16 output channels with 6-bit weights and activations. Each 6-bit minimum is divided into two 3-bit slices. A dual-mode digital-to-time converter (DM-DTC) encodes the most-significant slice in high-linearity (HL) mode and the least-significant slice in low-power (LP) mode. Readout is performed by a shared-clock time-to-digital converter (SC-TDC), in which one Gray-code time reference serves all paths while local latches preserve independent channel results. The training model reproduces code-dependent DTC nonlinearity, process–voltage–temperature variation, jitter, channel offset, TDC quantization, saturation, and scale mismatch. The architecture thereby combines significance-aware time encoding, channel-scalable readout, and hardware-aware adaptation. Post-layout simulations in 55 nm show that the 0.359 mm², 13.7 Kb design operates at 0.7–1.2 V and 5–30 MHz, consumes 0.025–0.324 mW, and achieves 43.2–94.3 TOPS/W. The normalized figure of merit is 6.01–13.09 POPS/W·bit². On CIFAR-10/ResNet-20, hardware errors reduce the baseline accuracy from 92.71% to 86.26%; error-aware training achieves 91.53%. Full article

Aviation ground safety and protective devices are critical for flight safety; however, their unintentional retention on aircraft after maintenance remains a persistent risk. Existing deep learning-based approaches for aviation safety have predominantly followed a reactive paradigm, detecting FOD on runways or inspecting the aircraft for inadvertently retained tools post-maintenance. In contrast, this paper advocates a proactive philosophy: using a neural network to recognize and inventory all ground safety and protective devices immediately after maintenance closure, thereby preventing retention incidents at their source. However, realizing this proactive verification is technically challenging—object detection for these devices often suffers from loss of fine-grained detail due to downsampling and inherently sparse semantic information of the targets. To this end, we propose an Information Retention and Feature Screening Synergistic Network (RS-Net) grounded in information bottleneck theory. The network comprises a main branch that enhances discriminative features through attention-guided screening, and an auxiliary branch, used only during training, that preserves fine-grained spatial details via information-retentive convolutions. A Dual-State Region Refinement Module (DRM) provides configurable support for both branches, decoupling the conflicting objectives of background compression and detail preservation. Experiments on a self-constructed dataset collected from real airline maintenance operations demonstrate that RS-Net substantially outperforms the strong YOLOv9 baseline, achieving gains of 4.531% in F1-score, 2.533% in mAP_0.5, and 1.429% in mAP_0.5:0.95. Cross-dataset experiments further validate its strong generalization capability. Full article