Accurate and efficient detection of road damage is essential for maintaining road safety and supporting intelligent transportation systems. While recent approaches leverage deep learning-based object detection frameworks, they often struggle with high computational demands and suboptimal feature extraction in complex environments. To address these challenges, we propose an enhanced object detection network for road damage detection based on the YOLOv8 architecture. Specifically, we integrate the Spatial and Channel Reconstruction Convolution (SCConv) module into the backbone to reduce feature redundancy while improving spatial and channel representation through a separation-reconstruction strategy. To enhance multi-scale feature fusion, we incorporate the Efficient Multi-Scale Attention (EMA) module into the neck, enabling adaptive spatial-channel attention without introducing significant computational overhead. Extensive benchmark comparison identifies YOLOv8l as a strong baseline for road damage detection. Building upon this, we embed SCConv within the original convolutional modules to achieve a lightweight network and enhance feature representation. We further explore multiple EMA integration strategies to identify an effective model configuration. Experimental results demonstrate that our best-performing model achieves higher detection accuracy than the baseline model, while maintaining computational efficiency.

Depth completion aims to predict dense depth from sparse sensor measurements with corresponding RGB guidance. Existing methods commonly suffer from texture copying and edge blurring due to the explicit fusion of RGB features. To address this, we propose a frequency-guided refinement approach that decouples structure from texture transfer in the frequency domain. Specifically, we decompose RGB features into wavelet sub-bands, and learn content-adaptive kernels that promote smooth propagation for low frequencies while preserving sharp boundaries for high frequencies. Importantly, RGB information serves only as conditioning signals to determine when and how filtering is applied, rather than being directly mixed with depth representations. To further improve robustness, a reliability-aware cross-stage modulation uses encoder features as priors to enhance trustworthy structures and suppress uncertain updates during multi-scale reconstruction. Extensive experiments on benchmark datasets demonstrate that our method generates high-fidelity depth maps with sharp edges and suppressed texture artifacts, achieving state-of-the-art performance.

Aim: This study aimed to validate and compare the diagnostic accuracy of a developed AI model against a periodontist, using an expert periodontist as the gold standard. The goal was to evaluate the AI's potential as a scalable triage tool for periodontal diagnosis in clinical settings.Materials and Methods: The AI model, developed from 2,000 panoramic radiographs, was clinically implemented using 300 anonymized images from the SIDEXIS program. It quantified alveolar bone loss to stage periodontitis, with diagnostic performance compared to a periodontist using expert evaluation as the reference.Results: The expert periodontist classified the cases as follows: 3.3% non-periodontitis, 10.3% stage I, 47.0% Stage II, 38.7% stage III, and 0.7% stage IV. The AI model demonstrated a higher overall consistency rate with the expert (71.7%) compared to the periodontist (62.7%). This superior performance was particularly notable in stage II (AI: 40% vs. periodontist: 28%) and stage III (AI: 29% vs. periodontist: 27%) classifications. The level of agreement with the expert, measured by Cohen's kappa, was moderate and similar for both the periodontist (kappa = 0.530) and the AI model (kappa = 0.497). The AI model exhibited significantly higher accuracy (97% vs. 87.3%) and perfect sensitivity (100% vs. 88.3%) but considerably lower specificity (10% vs. 60%) compared to the periodontist.Conclusions: The findings indicate AI's strong potential as a scalable triage tool to democratize periodontal care, delivering high diagnostic accuracy. With further refinement to improve specificity, such models could ensure equitable access to early intervention, overcoming geographic and socioeconomic barriers.

Cardiovascular disease (CVD) remains a major global health burden, motivating the development of accurate and scalable risk screening methods. This study presents a comparative evaluation of supervised machine learning models for CVD prediction using routinely collected clinical and lifestyle data. A unified machine learning pipeline was constructed, incorporating data preprocessing, feature transformation, and systematic model tuning. Multiple classifiers were evaluated under identical experimental settings, including logistic regression, support vector machine, multilayer perceptron, and ensemblebased tree models. Experimental results on a large public dataset show that ensemble methods consistently outperform classical classifiers. In particular, XGBoost achieves the best overall performance, with an accuracy of 87.02%, an AUC of 94.83%, and an F1-score of 8 5. 1 0%. These results highlight the potential of gradient-boosting-based machine learning models to support accurate and scalable cardiovascular disease risk screening using routinely collected health data.

Video deblurring faces a fundamental challenge, as blur degradation comprehensively affects frames by not only causing detail loss but also severely distorting structural information. This dual degradation across low- and high-frequency domains makes it challenging for existing methods to simultaneously restore both structural and detailed information through a unified approach. To address this issue, we propose a wavelet-based, blur-aware decoupled network (WBDNet) that innovatively decouples structure reconstruction from detail enhancement. Our method decomposes features into multiple frequency bands and employs specialized restoration strategies for different frequency domains. In the low-frequency domain, we construct a multi-scale feature pyramid with optical flow alignment. This enables accurate structure reconstruction through bottom-up progressive feature fusion. For high-frequency components, we combine deformable convolution with a blur-aware attention mechanism. This allows us to precisely extract and merge sharp details from multiple frames. Extensive experiments on benchmark datasets demonstrate the superior performance of our method, particularly in preserving structural integrity and detail fidelity.

This case report emphasizes the complexities involved in dental implant placement within the anterior esthetic zone, focusing on the integration of advanced surgical techniques and meticulous prosthetic design for optimal esthetic and functional results. A 28-year-old male presented with the absence of the upper left central incisor, which had been extracted 5 years prior due to fracture. Clinical and radiographic assessments indicated normal alveolar bone levels; however, the esthetic demands required a specialized approach. The surgical procedure included a papilla preservation flap, immediate implant placement, and bone grafting to prevent peri-implant resorption and achieve an optimal emergence profile. An Osstem implant and resorbable Xenograft were utilized to address the thin labial bone. Over 4 months, successful osseointegration occurred, followed by the placement of a permanent crown. This case illustrates that dental implant restoration in the anterior esthetic zone can effectively address functional and esthetic challenges, leading to high patient satisfaction with the final restoration’s natural appearance.

BackgroundPeriodontitis is a chronic inflammatory disease affecting the gingival tissues and supporting structures of the teeth, often leading to tooth loss. The condition begins with the accumulation of dental plaque, which initiates an immune response. Current radiographic methods for assessing alveolar bone loss are subjective, time-consuming, and labor-intensive. This study aims to develop an AI-driven model using Convolutional Neural Networks (CNNs) to accurately assess alveolar bone loss and provide individualized periodontal prognoses from panoramic radiographs.MethodsA total of 2,000 panoramic radiographs were collected using the same device, based on the periodontal diagnosis codes from the HOSxP Program. Image enhancement techniques were applied, and an AI model based on YOLOv8 was developed to segment teeth, identify the cemento-enamel junction (CEJ), and assess alveolar bone levels. The model quantified bone loss and classified prognoses for each tooth.ResultsThe teeth segmentation model achieved 97% accuracy, 90% sensitivity, 96% specificity, and an F1 score of 0.80. The CEJ and bone level segmentation model showed superior results with 98% accuracy, 100% sensitivity, 98% specificity, and an F1 score of 0.90. These findings confirm the models' effectiveness in analyzing panoramic radiographs for periodontal bone loss detection and prognostication.ConclusionThis AI model offers a state-of-the-art approach for assessing alveolar bone loss and predicting individualized periodontal prognoses. It provides a faster, more accurate, and less labor-intensive alternative to current methods, demonstrating its potential for improving periodontal diagnosis and patient outcomes.

BackgroundThe aging population is increasingly affected by periodontal disease, a condition often overlooked due to its asymptomatic nature. Despite its silent onset, periodontitis is linked to various systemic conditions, contributing to severe complications and a reduced quality of life. With over a billion people globally affected, periodontal diseases present a significant public health challenge. Current diagnostic methods, including clinical exams and radiographs, have limitations, emphasizing the need for more accurate detection methods. This study aims to develop AI-driven models to enhance diagnostic precision and consistency in detecting periodontal disease.MethodsWe analyzed 2,000 panoramic radiographs using image processing techniques. The YOLOv8 model segmented teeth, identified the cemento-enamel junction (CEJ), and quantified alveolar bone loss to assess stages of periodontitis.ResultsThe teeth segmentation model achieved an accuracy of 97%, while the CEJ and alveolar bone segmentation models reached 98%. The AI system demonstrated outstanding performance, with 94.4% accuracy and perfect sensitivity (100%), surpassing periodontists who achieved 91.1% accuracy and 90.6% sensitivity. General practitioners (GPs) benefitted from AI assistance, reaching 86.7% accuracy and 85.9% sensitivity, further improving diagnostic outcomes.ConclusionsThis study highlights that AI models can effectively detect periodontal bone loss from panoramic radiographs, outperforming current diagnostic methods. The integration of AI into periodontal care offers faster, more accurate, and comprehensive treatment, ultimately improving patient outcomes and alleviating healthcare burdens.

BACKGROUND:Artificial intelligence (AI) acts as the state-of-the-art in periodontitis diagnosis in dentistry. Current diagnostic challenges include errors due to a lack of experienced dentists, limited time for radiograph analysis, and mandatory reporting, impacting care quality, cost, and efficiency. OBJECTIVE:This review aims to evaluate the current and future trends in AI for diagnosing periodontitis. METHODS:A thorough literature review was conducted following PRISMA guidelines. We searched databases including PubMed, Scopus, Wiley Online Library, and ScienceDirect for studies published between January 2018 and December 2023. Keywords used in the search included "artificial intelligence," "panoramic radiograph," "periodontitis," "periodontal disease," and "diagnosis." RESULTS:The review included 12 studies from an initial 211 records. These studies used advanced models, particularly convolutional neural networks (CNNs), demonstrating accuracy rates for periodontal bone loss detection ranging from 0.76 to 0.98. Methodologies included deep learning hybrid methods, automated identification systems, and machine learning classifiers, enhancing diagnostic precision and efficiency. CONCLUSIONS:Integrating AI innovations in periodontitis diagnosis enhances diagnostic accuracy and efficiency, providing a robust alternative to conventional methods. These technologies offer quicker, less labor-intensive, and more precise alternatives to classical approaches. Future research should focus on improving AI model reliability and generalizability to ensure widespread clinical adoption.

Necrotizing fasciitis (NF) is a rare but life-threatening soft tissue infection. Early diagnosis and treatment are crucial to prevent complications such as amputations and even death. However, distinguishing NF from cellulitis, a less severe skin infection, can be difficult. Machine learning (ML) has the potential to improve this distinction. Three algorithms for classification are logistic regression, decision tree, and random forests. The dataset was collected from Maha Sarakham Hospital, Thailand. It includes six features: creatinine, hemoglobin, platelet, sodium, white blood cell, and C-reactive protein. The target results consist of two classes (NF and non-NF). A trained model will be tested on a separate dataset to evaluate its performance. The C-reactive protein data has some missing values. The experiment has two cases: using all instances with missing data and removing records with missing data. Metrics such as accuracy, precision, sensitivity, specificity, and F-measure will be used. The study indicates that random forests perform well in all instances with missing data, while the decision tree is suitable for small datasets.

Necrotizing fasciitis (NF) is a severe and life-threatening soft tissue infection that requires timely and accurate diagnosis to improve patient outcomes. The early diagnosis of NF remains challenging due to its similarity to other subcutaneous soft tissue infections like cellulitis. This study aims to employ machine learning techniques to differentiate NF from cellulitis and enhance the diagnostic accuracy of NF by developing a modified LRINEC (MLRINEC) score. These modifications aimed to improve the sensitivity and specificity of NF diagnosis. The study utilized three machine learning classifiers—Logistic Regression, decision tree, and Random Forest—to assess their effectiveness in distinguishing between NF and cellulitis cases. The MLRINEC score was developed by incorporating six key blood test parameters: creatinine, hemoglobin, platelet count, sodium, white blood cell count, and C-reactive protein using laboratory data from Maha Sarakham Hospital in Northeastern Thailand. Our findings indicate that the decision tree classifier demonstrated superior performance, achieving the highest recall, particularly in accurately identifying NF cases. A feature importance analysis revealed that hemoglobin levels and white blood cell counts were the most critical factors influencing the model’s predictions. The platelet count (PT), C-reactive protein (CRP), and creatinine (CT) also played important roles, while sodium levels (NA) were the least influential. The MLRINEC score demonstrates high accuracy in classifying NF and cellulitis patients, paving the way for improved diagnostic protocols in clinical settings.

Aim & Background: Periodontal diseases are a significant public health concern, affecting over a billion people worldwide. Conventional diagnostic and prognostic methods, which rely on clinical examination and radiographic interpretation for each tooth, often lack consistency and are labor-intensive and time-consuming. This highlights the need for more reliable and efficient tools to improve these processes. This study aims to develop an innovative AI-driven model for analyzing panoramic radiographs to detect alveolar bone loss and determine periodontal prognostication. Materials and methods: A balanced dataset of 2,000 panoramic radiographs, equally representing periodontally healthy individuals and those with periodontitis, was collected. Image enhancement techniques were applied, and an AI model using YOLOv8 was developed to segment teeth, identify the cemento-enamel junction (CEJ), and assess alveolar bone levels to quantify bone loss and classify the prognosis for each tooth. Results: The teeth segmentation model achieved precision, recall, and F1 scores of 0.8, 0.9, and 0.8, respectively. The CEJ and alveolar bone level segmentation model demonstrated high performance, with scores of 0.9, 1.0, and 0.9. These results highlight the model’s capability to accurately identify critical features of periodontal disease, reducing diagnostic variability and enhancing prognostic assessments. Conclusion: This AI model sets a new benchmark for detecting alveolar bone loss and determining periodontal prognostication. The developed AI model offers a promising solution by providing quicker, less labor-intensive, and more precise alternatives to current diagnostic approaches. Clinical Significance: The model’s ability to provide consistent, objective assessments can significantly reduce the workload of dental professionals, ensuring better patient outcomes.

With the rapid development of computer technology, the intelligent driving of automobiles has become a popular research field. Road damage detection is critical in intelligent automobile driving and has been developed for a long time. The early detection method is to detect by embedding several sensors in the car. In recent years, deep learning methods have also been gradually applied in the research of pavement damage detection. This paper mainly takes road damage detection based on the YOLOv8 by comparing it to YOLOv5 and multiple pre-train models of YOLOv8. This paper describes the solution using YOLO to detect the various types of road damage in the Crowdsensing-based Road Damage Detection Challenge (CRDDC'2022). The dataset is separated into training, validation, and test sets. The medium and large pre-train models of YOLOv8 have the highest mAP and F1-score at 0.62 and 0.61, respectively. YOLOv8 of the small and medium pre-train model provides better performance than YOLOv5. For comparing YOLOv8, the performance of the large pre-train model is better than the others, but it is not much different from the medium pre-train model. For this reason, the medium pre-train model may be suitable for the real-time problem as it does not take time for training.

Currently, lime is a type of plant that has been cultivated in many places. Because limes are used in cooking, and their properties are herb. In Thailand, limes are used for drinking and medical herb to used in health care. From this popularity, the cultivation of limes became widespread and began to be planted more in farm and more at home. Nowadays, lime cultivation can be controlled to produce off-season produce. However, lime is also prone to diseases without proper care. For these reasons, this research is to study the analyzation of decease from lime leaves by using deep learning. The Convolutional Neural Networks (CNNs) of deep learning is used to classify the decease. The proposed architecture of CNNs in this study is to compare to LeNet-5, VGG16, and RestNet-50 architectures. The total number of single lime leaf images is 5,710. The input images are RGB color. The normal and decease lime leaves are separated in equal. Training and test sets are 80 and 20 percent, respectively. The evaluation result is found that LeNet-5 has the lowest accuracy, while the proposed architecture has the highest accuracy but it is not different from ResNet-50.

Handwritten recognition is a difficult task. The conventional technique relies on the character segmentation, feature extraction, and classification process. The segmentation is a tremendous challenge when there are variation of character patterns and alignments in a sentence, such as linking segments between characters in the Thai language. Promising segmentation outcome is favorable but not applicable in most applications. This work proposes a methodology for Thai handwritten recognition by applying Convolutional Neural Networks (CNNs) and Recurrent Neural Networks (RNNs). The first step is text localization before feeding to the network. CNN extracts the abstract features before they are fed to RNN to learn the sequence of characters in an image. The optimization is performed with an integrated Connectionist Temporal Classification (CTC) module (to arrange the final results). A standard Thai handwritten dataset (BEST2019) and more collection are used in this study. for training and test sets. The experimental results show that the integration of CNN and RNN provides promising results of the test set with a Character Error Rate (CER) of 1.58%. For testing with the seen and unseen dataset of the final round of BEST2019 competition, the CER is at 24.53%.

This paper proposes the improvement performance of Sparse Auto Encoder (SAE) by using Dominant Patterns of Histograms of Oriented Gradients (DPHOG). The SAE has a simple structure and fast computation. However, the SAE demonstrates less accuracy than the Convolution of Neuron Networks (CNN). This proposed method selects the dominant features of female and male face images, and then encoding and decoding these features within SAE. The researchers conducted DPHOG with SAE in two datasets and compared them with SAE, HOG with SAE, and CNN. Experimental results showed that DPHOG with SAE produced the highest performance in terms of accuracy, true positive rate, false positive rate and time computation in the two datasets. In addition, the DPHOG with SAE can extract learning features the form small dataset. In contrast, CNN requires a huge dataset for learning features.

The word ‘encoded’ has been updated to ‘decoded’ in section 4.3 of the chapter.

Automatic transcription of ancient handwritten manuscripts can be a challenging task when compared with a transcription of contemporary handwriting. Characters and words can have unusual and varying shapes, with significant variation between writers, and sufficient labelled data from which to train machine learning algorithms can be difficult to access. This paper describes ancient Thai handwriting transcription on block-based from archive manuscripts, using a hybrid deep neural network with both convolutional (CNN) and recurrent (RNN) layers, trained using Connectionist Temporal Classification (CTC) loss. Six architecture variations are compared. Data augmentation was applied to synthetically increase the number of training samples, resulting in improved learning. Thai archive manuscripts were collected from the Thai National Library. The character error rate (CER) in the best architecture was found to be 11.9 percent.

This paper aims to apply a YOLOv3 technique for detecting and recognizing Thai traffic signs in real-time environments. The Thai Traffic Sign Dataset (TTSD) was collected by car cameras to store the video images using the resolution of 1920 × 1080 pixels using 60 frames per second, and a 1280 × 720 pixels and 30 frames per second. In addition, the data was collected in the rural area of Maha Sarakham Province and Kalasin Province. The dataset was generated and distributed for general traffic sign detection and recognition. Two architectures (YOLOv3 and YOLOv3 Tiny) are compared with 50 classes of road signs and 200 badges in each class, containing 9,357 images. The experiment shows that the mean average precision (mAP) of YOLOv3 (88.10

Today, speech-based information retrieval is necessary because of the various keyboard less applications such as user-friendly interface for PCs, car navigation systems, or mobile-phones. Unfortunately, current programs may not support for every language such as Thai language. Therefore, the goal of this work is to develop a system to retrieve documents by Thai speech. The proposed prototype application provides for desktop device such personal computer (PC). This application will accept Thai speech queries and then, it provides Thai text search result on the PC screen.