Augmentation Techniques Research Articles

Application of artificial intelligence-based detection of furcation involvement in mandibular first molar using cone beam tomography images- a preliminary study

BackgroundRadiographs play a key role in diagnosis of periodontal diseases. Deep learning models have been explored for image analysis in periodontal diseases. However, there is lacuna of research in the deep learning model-based detection of furcation involvements [FI]. The objective of this study was to determine the accuracy of deep learning model in the detection of FI in axial CBCT images.MethodologyWe obtained initial dataset 285 axial CBCT images among which 143 were normal (without FI) and 142 were abnormal (with FI). Data augmentation technique was used to create 600(300 normal and 300 abnormal) images by using 200 images from the training dataset. Remaining 85(43 normal and 42 abnormal) images were kept for testing of model. ResNet101V2 with transfer learning was used employed for the analysis of images.ResultsTraining accuracy of model is 98%, valid accuracy is 97% and test accuracy is 91%. The precision and F1 score were 0.98 and 0.98 respectively. The Area under curve (AUC) was reported at 0.98. The test loss was reported at 0.2170.ConclusionThe deep learning model (ResNet101V2) can accurately detect the FI in axial CBCT images. However, since our study was preliminary in nature and carried out with relatively smaller dataset, a study with larger dataset will further confirm the accuracy of deep learning models.

Enhanced Panoramic Radiograph-Based Tooth Segmentation and Identification Using an Attention Gate-Based Encoder–Decoder Network

Background: Dental disorders are one of the most important health problems, affecting billions of people all over the world. Early diagnosis is important for effective treatment planning. Precise dental disease segmentation requires reliable tooth numbering, which may be prone to errors if performed manually. These steps can be automated using artificial intelligence, which may provide fast and accurate results. Among the AI methodologies, deep learning has recently shown excellent performance in dental image processing, allowing effective tooth segmentation and numbering. Methods: This paper proposes the Squeeze and Excitation Inception Block-based Encoder–Decoder (SE-IB-ED) network for teeth segmentation in panoramic X-ray images. It combines the InceptionV3 model for encoding with a custom decoder for feature integration and segmentation, using pointwise convolution and an attention mechanism. A dataset of 313 panoramic radiographs from private clinics was annotated using the Fédération Dentaire Internationale (FDI) system. PSPL and SAM augmented the annotation precision and effectiveness, with SAM automating teeth labeling and subsequently applying manual corrections. Results: The proposed SE-IB-ED network was trained and tested using 80% training and 20% testing of the dataset, respectively. Data augmentation techniques were employed during training. It outperformed the state-of-the-art models with a very high F1-score of 92.65%, mIoU of 86.38%, and 92.84% in terms of accuracy, precision of 92.49%, and recall of 99.92% in the segmentation of teeth. Conclusions: According to the results obtained, the proposed method has great potential for the accurate segmentation of all teeth regions and backgrounds in panoramic X-ray images.

Architecture-Aware Augmentation: A Hybrid Deep Learning and Machine Learning Approach for Enhanced Parkinson’s Disease Detection

Parkinson’s Disease (PD) is a progressive neurodegenerative disorder affecting millions worldwide. Early detection is crucial for improving patient outcomes. Spiral drawing analysis has emerged as a non-invasive tool to detect early motor impairments associated with PD. This study examines the performance of hybrid deep learning and machine learning models in detecting PD using spiral drawings, with a focus on the impact of data augmentation techniques. We compare the accuracy of Vision Transformer (ViT) with K-Nearest Neighbors (KNN), Convolutional Neural Networks (CNN) with Support Vector Machines (SVM), and Residual Neural Networks (ResNet-50) with Logistic Regression, evaluating their performance on both augmented and non-augmented data. Our findings reveal that ViT with KNN, initially achieving 96.77% accuracy on unaugmented data, experienced a notable decline across all augmentation techniques, suggesting it relies heavily on global patterns in spiral drawings. In contrast, ResNet-50 with Logistic Regression showed consistent improvement with data augmentation, reaching 93.55% accuracy when rotation and flipping techniques were applied. These results highlight that hybrid models respond differently to augmentation, and careful selection of augmentation strategies is necessary for optimizing model performance. Our study provides important insights into the development of reliable diagnostic tools for early PD detection, emphasizing the need for appropriate augmentation techniques in medical image analysis.

Retinal Image Augmentation using Composed GANs

Medical image analysis faces a significant challenge in the scarcity of annotated data, which is crucial for developing generalizable Deep Learning (DL) models that require extensive training data. Consequently, the field of medical image generation has garnered substantial interest and potential for further exploration. Besides widely employed data augmentation techniques, such as rotation, reflection, and scaling, Generative Adversarial Networks (GANs) have demonstrated the ability to effectively leverage additional information from datasets by generating synthetic samples from real images. In the context of retinal image synthesis, an image-to-image translation approach is frequently adopted to generate retinal images from available vessel maps, which can be scarce and resource-intensive to obtain. Deviating from prior work reliant on pre-existing vessel maps, this study proposes a learning-based model that is independent of vessel maps, utilizing Progressive Growing GAN (PGGAN) to generate vascular networks from random noise. The visual and quantitative evaluations conducted suggest that the majority of the images generated by the proposed model are substantially distinct from the training set while maintaining a high proportion of true image quality, underscoring the model's potential as a powerful tool for data augmentation.

An unmanned aerial vehicle captured dataset for railroad segmentation and obstacle detection

Safety is crucial in the railway industry because railways transport millions of passengers and employees daily, making it paramount to prevent injuries and fatalities. In order to guarantee passenger safety, computer vision, unmanned aerial vehicles (UAV), and artificial intelligence will be essential tools in the near future for routinely evaluating the railway environment. An unmanned aerial vehicle captured dataset for railroad segmentation and obstacle detection (UAV-RSOD) comprises high-resolution images captured by UAVs over various obstacles within railroad scenes, enabling automatic railroad extraction and obstacle detection. The dataset includes 315 raw images, along with 630 labeled and 630 masked images for railroad semantic segmentation. The dataset consists of 315 original images captured by the UAV for object detection and obstacle detection. To increase dataset diversity for training purposes, we applied data augmentation techniques, which expanded the dataset to 2002 augmented and annotated images for obstacle detection cover six different classes of obstacles on railroad lines. Additionally, we provide the original 315 images along with a script for augmentation, allowing users to generate their own augmented data as needed, offering a more sustainable and customizable option. Each image in the dataset is accurately annotated with bounding boxes and labeled under six categories, including person, boulder, barrel, branch, jerry can, and iron rod. This comprehensive classification and detailed annotation make the dataset an essential tool for researchers and developers working on computer vision applications in the railroad domain.

Enhancing camouflaged object detection through contrastive learning and data augmentation techniques

Camouflaged object detection (COD) aims to locate and segment objects that blend into their surroundings, presenting significant challenges due to the high similarity between the objects and their background. This work introduces a novel approach, Contrastive Learning with Augmented Data (CLAD), which enhances COD performance by leveraging contrastive learning and data augmentation. Our method formulates a simplified task by placing camouflaged objects in new environments, creating positive and negative samples for contrast learning. This process strengthens the model’s ability to differentiate camouflaged objects from complex backgrounds. Furthermore, we introduce a concatenated feature enhancement module to integrate and enrich multi-scale features, improving the overall expressive power of the model. Extensive experiments on four benchmark datasets demonstrate that CLAD outperforms state-of-the-art COD methods, and its effectiveness extends to salient object detection tasks, achieving competitive results across multiple metrics.

Deep Learning for CAD Prediction: X-ray Angiography Insights

This study presents a deep learning-based approach to improve the prediction of coronary artery disease (CAD) using X-ray angiography images. The primary objective is to achieve accurate and automated CAD identification by employing a convolutional neural network (CNN) model. The methodology involves preprocessing the dataset through normalization and augmentation techniques and utilizes a U-Net architecture for precise detection of coronary stenosis. To ensure robustness and generalizability, hyperparameter tuning and dropout regularisation are applied during model training. The proposed model achieves high performance, with an average Dice coefficient of 0.57 and a Jaccard Index of 0.47 on a held-out test set, indicating its effectiveness in segmenting coronary artery stenosis. These findings support the integration of deep learning methods into clinical workflows for enhanced CAD diagnosis and early intervention.

Evaluation of post-tetanic motor evoked potential as an augmentation technique under partial neuromuscular blockade during craniotomy

ObjectiveIn craniotomies requiring motor evoked potential (MEP) monitoring, avoiding neuromuscular blockade (NMB) is preferable, but its complete avoidance poses risks of unexpected movement. This retrospective study investigates the application of a post-tetanic MEP augmentation technique to enhance baseline recording of transcranial stimulation MEP (Tc-MEP) under partial NMB during craniotomy. MethodsTwenty-six patients were included. The level of partial NMB was maintained at a train-of-four ratio of approximately 40 %. Monophasic constant-current stimulation was applied on the craniotomy side with + 20 % of the threshold intensity. Post-tetanic Tc-MEP, involving tetanic stimulation of the median nerve 1 s before transcranial stimulation, was performed on patients who failed to record using conventional baseline recording. ResultsThe post-tetanic Tc-MEP technique successfully improved the success rate of baseline recording from 61.5 % to 100 %. Application of post-tetanic Tc-MEP significantly increased amplitudes in both the upper (p = 0.04) and lower limbs (p < 0.01) compared to before post-tetanic Tc-MEP. No patients had unexpected movements. ConclusionsThis study indicates that post-tetanic Tc-MEP enhanced the success rate of baseline recording during craniotomy under partial NMB. SignificanceThe combination of partial NMB and post-tetanic Tc-MEP could be a useful regimen for craniotomy with MEP monitoring, addressing both safety concerns and successful baseline recording.

Stacking-based ensemble learning for identifying artist signatures on paintings

Identifying artist signatures on paintings is essential for authenticating artworks and advancing digital humanities. An artist’s signature is a consistent element included in each painting that the artist creates, providing a unique identifier for their work. Traditional methods that rely on expert analysis and manual comparison are time-consuming and are prone to human error. Although convolutional neural networks (CNNs) have shown promise in automating this process, existing single-model approaches struggle with the diversity and complexity of artistic styles, leading to limitations in their performance and generalizability. Therefore, this study proposes an ensemble learning approach that integrates the predictive power of multiple CNN-based models. The proposed framework leverages the strengths of three state-of-the-art CNNs: EfficientNetB4, ResNet-50, and Xception. These models were independently trained, and the predictions were combined using a meta-learning strategy. To address class imbalance, data augmentation techniques and weighted loss functions were employed. The experimental results obtained on a dataset of more than 8,000 paintings from 50 artists demonstrate significant improvements over individual CNN architectures and other ensemble methods, thereby effectively capturing complex features and improving generalizability.

Deep learning utilization in Sundanese script recognition for cultural preservation

This study addresses the challenge of preserving the Sundanese script, a traditional writing system of the Sundanese community in Indonesia, which is at risk of being forgotten due to technological advancements. To tackle this problem, we propose a deep learning approach using the YOLOv8 model for the automatic recognition of Sundanese characters. Our methodology includes creating a comprehensive dataset, applying augmentation techniques, and annotating the characters. The trained model achieved a precision of 95% after 150 epochs, demonstrating its effectiveness in recognizing Sundanese characters. While some variability in accuracy was observed for certain characters and real-time applications, the results indicate the feasibility and promise of using deep learning for Sundanese script recognition. This research highlights the potential of technological solutions to digitize and preserve the Sundanese script, ensuring its continued legacy and accessibility for future generations. Thus, we contribute to cultural preservation by providing a method to safeguard the Sundanese script against obsolescence.

Advanced montane bird monitoring using self-supervised learning and transformer on passive acoustic data

Passive acoustic monitoring combined with deep learning-based bird sound classifiers is an effective tool, particularly in remote areas. While self-supervised learning has recently excelled in natural language processing and image recognition, its application to bird sound recognition remains limited. This study proposes an innovative self-supervised learning approach, which leverages vast amounts of passive acoustic recordings for pre-training, followed by fine-tuning of target species. Compared to the three state-of-the-art models based on transfer learning from ImageNet, the proposed method demonstrated improvements across all species, with even more significant gains for tail-end species. These results confirm that domain-specific pre-training in self-supervised learning enhances downstream recognition tasks and provides greater robustness, benefiting tail-end species in imbalanced ecological datasets. Our experiments further demonstrate that integrating open-source datasets and data augmentation techniques is the most effective strategy for mitigating data imbalances and cross-domain issues. In addition, introducing a ‘catch-all’ category into training datasets has been shown to improve model robustness in open set recognition scenarios. We also identified the minimum viable sample size requirements for our proposed model and explored the impact of overlapping bird vocalizations during dawn choruses on model performance. Targeting 31 bird species in the montane regions of subtropical Taiwan, the model achieved a class-wise mean average precision of 0.782 and an overall precision of 85.6 % at the F0.5 threshold in dawn chorus soundscape recordings. This study confirms the effectiveness and advantages of self-supervised learning in bird sound recognition, supporting long-term monitoring of bird distribution and vocal activity in remote montane areas.

The Open-Book Technique With Sliced Biceps Tendon Autograft Augmentation in the Reconstruction of the Subscapularis in Shoulder Arthroplasty

Postoperative insufficiency of the subscapularis can diminish clinical outcome after shoulder arthroplasty. Poor tendon tissue quality and prior surgery can impair the possibility of reconstruction and healing, as well as the postoperative range of motion and strength. Patch augmentation has been investigated for soft tissue augmentation in the glenohumeral joint utilizing different graft materials. Recently, the surgical technique using compressed biceps tendon autograft was introduced for the subscapularis augmentation following shoulder arthroplasty. However, the viable cell potential after mechanical compression remains uncertain. Therefore, we propose an augmentation technique of the subscapularis tendon using a sliced autograft of the long head of the biceps tendon. This technique can offer a simple, replicable, and cost-effective autologous tendon patch augmentation after subscapularis tenotomy in the setting of total shoulder arthroplasty.

Uni-to-Multi Modal Knowledge Distillation for Bidirectional LiDAR-Camera Semantic Segmentation.

Combining LiDAR points and images for robust semantic segmentation has shown great potential. However, the heterogeneity between the two modalities (e.g. the density, the field of view) poses challenges in establishing a bijective mapping between each point and pixel. This modality alignment problem introduces new challenges in network design and data processing for cross-modal methods. Specifically, 1) points that are projected outside the image planes; 2) the complexity of maintaining geometric consistency limits the deployment of many data augmentation techniques. To address these challenges, we propose a cross-modal knowledge imputation and transition approach. First, we introduce a bidirectional feature fusion strategy that imputes missing image features and performs cross-modal fusion simultaneously. This allows us to generate reliable predictions even when images are missing. Second, we propose a Uni-to-Multi modal Knowledge Distillation (U2MKD) framework, leveraging the transfer of informative features from a single-modality teacher to a cross-modality student. This overcomes the issues of augmentation misalignment and enables us to train the student effectively. Extensive experiments on the nuScenes, Waymo, and SemanticKITTI datasets demonstrate the effectiveness of our approach. Notably, our method achieves an 8.3 mIoU gain over the LiDAR-only baseline on the nuScenes validation set and achieves state-of-the-art performance on the three datasets.

MPCNN: a novel approach for detecting human Monkeypox from skin lesion images leveraging deep neural network

The global healthcare scenario encounters a substantial challenge caused by the widespread outbreak of Monkeypox affecting over 65 countries. Limited availability of polymerase chain reaction (PCR) tests and biochemical assays necessitates alternative strategies. This study explores the viability of computer-aided identification of Monkeypox through the analysis of skin lesion images, offering a potential solution, particularly in resource-constrained settings. Employing data augmentation techniques, we augment the dataset to enhance its robustness. Subsequently, we utilize various pre-trained deep learning models, including EfficientNetB3, VGG16, ResNet50, AlexNet, and EfficientNet for classification tasks related to Monkeypox and other diseases. The achieved accuracies for these models are 98.48%, 69.19%, 91.41%, 78.38%, and 94.44%, respectively. We introduce a novel modified convolutional neural network (CNN) architecture named MPCNN to further improve performance. Our proposed MPCNN model demonstrates exceptional accuracy, precisely identifying Monkeypox patients with a remarkable precision of 99.49%. This technological advancement in disease identification holds significant promise for enhancing healthcare strategies and response mechanisms in the context of global health concerns.

Fog and rain augmentation for license plate recognition in tropical country environment

Automatic license plate recognition (ALPR) is a critical component in modern traffic management systems. However, ALPR systems often face challenges in accurately recognizing license plates under adverse weather conditions, such as fog and rain, prevalent in tropical regions. Deep learning ALPR models necessitate huge and diverse datasets for robustness, but data availability remains a concern since unpredictable fog and rain patterns hinder data collection. In this study, we address the issue of enhancing ALPR's robustness by introducing a novel augmentation strategy that combines traditional and weather augmentation techniques. By augmenting the dataset with weather-induced variations, we aim to improve the generalization capability of ALPR models, enabling them to handle a wider range of weather-related challenges. We also investigate the synergy between these weather augmentations and established scene text recognition (STR) methods, such as convolutional recurrent neural network (CRNN), TPS-ResNet BiLSTM-attention (TRBA), autonomous bidirectional iterative scene text recognition (ABINet), vision transformer (ViTSTR), and permutated autoregressive sequence (PARSeq), to determine their impact on recognition accuracy. Experiments using different training data sets show that training data containing a combination of traditional and weather augmentation produces the best accuracy and 1-NED performance compared to training data without augmentation and traditional augmentation only. The average increase accuracy of all STR model is 1.13% with the best increase accuracy of 3.68% using TRBA.

Ensemble transfer learning meets explainable AI: A deep learning approach for leaf disease detection

Global food security is threatened by plant diseases and manual detection methods are often labor-intensive and time-consuming. Deep learning offers a promising solution by enabling early and accurate detection of leaf diseases. This study presents a novel deep-learning model designed to address the challenges of real-world leaf disease identification. To enhance the model's robustness, we incorporated six datasets (LD, LD1, LD2, LD3, LD4, LD5) which include image augmentation techniques, like flipped versions (LD1) and controlled noise (LD2, LD3). Additionally, we introduced new datasets with additional noise types (LD4) and real-world scenarios (LD5). To further improve accuracy, we employed an ensemble approach, combining MobileNetV3_Small and EfficientNetV2B3 with weighted voting. Our model achieved exceptional performance, surpassing 94 % accuracy on imbalanced data (LD) and exceeding 99 % on balanced, high-quality data (LD1). Even in noisy environments (LD2, LD3, LD4, LD5), our model consistently outperformed other approaches, maintaining an accuracy rate above 90 %. To ensure transparency and interpretability, we utilized Explainable AI (LIME) to visualize the model's decision-making process. These results demonstrate the potential of our model as a reliable and accurate tool for leaf disease detection in practical agricultural settings.

MobileNet Backbone Based Approach for Quality Classification of Straw Mushrooms (Volvariella volvacea) Using Convolutional Neural Networks (CNN)

Straw mushrooms (Volvariella volvacea) are a crucial commodity in Indonesia, with consumption on the rise due to their nutritional value and increasing demand for healthy food options. Despite this growth, farmers often struggle with accurately assessing the post-harvest quality of mushrooms according to market standards, which can diminish their economic value. Manual classification, which relies on human judgment and estimation, is frequently inefficient and susceptible to errors such as inconsistencies in quality assessment and limitations in detecting subtle variations. This study aims to automate the classification of straw mushrooms based on quality using deep learning, specifically by employing MobileNetv3 as the backbone for classifying mushrooms based on their shape and color by the Indonesian National Standards (SNI). The MobileNet-CNN Backbone model implemented in this study demonstrated exceptional performance, achieving a classification accuracy of 99%, thus proving its effectiveness and reliability in replacing traditional manual methods. The results of this research indicate significant potential for applying deep learning models to enhance the efficiency and precision of mushroom quality assessment. However, there remain challenges that require further development, including adding more diverse background data, improving image resolution, and refining data augmentation techniques. Addressing these challenges is essential for achieving optimal results in varying environmental conditions, ensuring the model can be broadly implemented in the agricultural industry. Such advancements could lead to more consistent and accurate quality assessments, benefiting producers and consumers in the mushroom market.

Contrastive learning-based multi-view clustering for incomplete multivariate time series

Incomplete multivariate time series (MTS) clustering is a prevalent research topic in time series analysis, aimed at partitioning MTS containing missing data into distinct clusters. Contrastive learning-based multi-view clustering methods are a promising approach to address this issue. However, existing methods are typically not designed for time series. Specifically, most of these methods struggle to capture the inherent properties of time series, and are susceptible to losing their interdimensional correlations, thereby compromising data integrity. Furthermore, they commonly utilize data augmentation techniques to generate sample pairs for contrastive learning. These existing data augmentation techniques are not suitable for time series, and introduce uncertainty factors, which can diminish the representation learning capacity of contrastive learning. To address the challenges, we propose a contrastive learning-based multi-view clustering method for incomplete multivariate time series (MVCIMTS). In this method, each variable within the MTS is treated as a separate view, enabling a multi-view learning approach. To better leverage the intrinsic information of time series, we utilize a GRU-based model architecture that integrates imputation and clustering within a unified deep learning framework. In this way, missing views can be effectively inferred, and representations suitable for clustering can be learned, thereby enhancing the clustering performance for incomplete time series. Furthermore, we introduce an innovative contrastive learning approach specifically tailored for MTS, which ensures that the exploration of common semantics and clustering consistency across views remains unaffected by uncertainty factors. It assumes that each time series variable within the same sample has similar representations, thereby taking into account the correlation between variables and enhancing the quality of the representations. To the best of our knowledge, this is the first attempt at applying contrastive learning-based multi-view deep clustering to incomplete MTS. We conduct extensive comparative experiments with five multi-view clustering methods and two time series clustering methods on seven benchmark datasets. The results demonstrate that our proposed method is superior to other state-of-the-art methods.

EARLY DETECTION OF BRAIN TUMOR USING MRI AND TRANSFER LEARNING

Brain tumors pose significant risks to cognitive functions, making early detection crucial for improving patient survival rates. Accurate tumor detection aids neuro-oncologists in diagnosing tumor types and recommending appropriate treatments. However, manual detection is challenging, time-consuming, and prone to human error. Recently, Deep Learning (DL) models have demonstrated substantial potential in efficiently classifying large image datasets. This paper presents three novel and efficient multi-class DL architectures leveraging transfer learning to classify different brain tumors. Our primary contribution is to enhance the predictive accuracy while minimizing the usage of computational resource compared to other state-of-art models in the literature by forgoing preprocessing, segmentation, hybrid models, and augmentation techniques. Additionally, we reduce the number of FC layers to streamline computation. We conduct extensive experiments to evaluate the performance of our models using the brain tumor Figshare T1-weighted contrast-enhanced MRI dataset, comprising 3064 images from three distinct tumor types. The InceptionV3 model records a 98.36% accuracy level with five-fold cross-validation. By incorporating batch normalization and optimizing the learning rate for the Adam optimizer, the Xception model reached 99.19% accuracy. Finally, we utilize Particle Swarm Optimization (PSO) to fine-tune the learning rate of the Stochastic Gradient Descent (SGD) optimizer. The Xception model attained 98.85% accuracy. These results highlight the novelty of our approach, offering a practical solution for neuro-oncologists, particularly through the fine-tuned Xception model, which delivers early and accurate tumor detection with minimal computational resources.

Bone-marrow-based augmentation in rotator cuff repair—from microfracture to bone marrow aspirate concentrate

AbstractRotator cuff tears are a major cause of shoulder dysfunction, and relatively high rates of rotator cuff re-tear persist despite surgical advancements, particularly in patients with poor biological healing potential. Bone-marrow-based augmentation techniques, specifically bone marrow stimulation (BMS) and bone marrow aspirate concentrate (BMAC), have been introduced to enhance the biological environment at the repair site, potentially improving outcomes. Bone marrow stimulation, commonly achieved through microfracture, stimulates the release of growth factors and mesenchymal stem cells (MSCs) from the bone marrow to promote tendon-to-bone integration. Although simple and cost-effective, clinical results for BMS augmentation in rotator cuff repair (RCR) are mixed, with most recent meta-analyses not demonstrating a clinically significant superiority over conventional RCR. Augmentation with BMAC offers a more sophisticated approach, concentrating MSCs and anti-inflammatory cytokines to directly enhance the healing process. Preclinical studies have shown promising results with BMAC augmentation, demonstrating improved tendon integrity and biomechanical strength. The existing clinical studies suggest BMAC may reduce re-tear rates and enhance tendon healing, although outcomes are not yet universally optimal. The current evidence highlights the potential of these techniques, particularly as a potential treatment option in biologically challenging cases. However, the variability in clinical outcomes underscores the need for further research to refine these methods and establish their role in routine clinical practice.