Search

The teeth are one of the essential organs in the human digestive system. Teeth not only help people in the digestive process but also have a vital function for human-sense. Therefore, it is necessary to maintain dental health to avoid caries, the most common dental disease in developed countries. The most recent method to detect caries is using an X-ray machine. However, as technology has developed and advanced, diagnosing dental caries images can be performed using artificial intelligence and image processing. This study utilized four classes (1, 2, 3 and 4) of dental caries images. The raw images encompassed 347 dental caries images, divided into 314 training images and 33 testing images. This study aims to determine the feature extraction process of the Zernike Moment Invariant method and the process and results of dental caries classification with the input of extraction results classified using SVM and KNN. The classification results represented the accuracy of each SVM and KNN model. The highest accuracy of 71.9% was acquired from KNN. [ABSTRACT FROM AUTHOR]

The classification of vehicle images, such as cars, using industrial automation technology can aid the police, particularly in traffic analysis. It is possible to determine cars' brand, model, and year of manufacture through their shape images. This study aims to design and analyze a system that can perform feature extraction and classification of car classes based on shape using a combined method of Histogram of Oriented Gradients 1 (HOG1) and Histogram of Oriented Gradients 2 (HOG2), Support Vector Machine (SVM), and K-Nearest Neighbor (KNN) on the Matlab R2020a application. This study deployed car front view images, with 40 training images after pre-processing (12 Dodge Charger SRT-8 2009, 8 Dodge Dakota Crew Cab 2010, 10 Hyundai Elantra Touring Hatchback 2012, and 10 Spyker C8 Coupe 2009) and 10 testing images after pre-processing (4 Dodge Charger SRT-8 2009, 2 Dodge Dakota Crew Cab 2010, 2 Hyundai Elantra Touring Hatchback 2012, 2 Spyker C8 Coupe 2009). The research findings with the combined feature extraction of HOG1 and HOG2 unveiled that the Quadratic SVM classification yielded the highest SVM accuracy, with a maximum accuracy of 97.5% for training data and 100% for testing data. Meanwhile, the fine KNN classification, with a maximum accuracy of 95% for training data and 60% for testing data, generated the greatest KNN accuracy. [ABSTRACT FROM AUTHOR]

Nowadays, there are an increasing number of vehicles that have developed, resulting in different shapes. Therefore, the variety of vehicle types in traffic flow needs to be monitored by a classification system built-in in machine learning. This vehicle classification research utilized four classes of the vehicle dataset: class 3 "Acura TL Sedan 2012", class 14 "Audi TTS Coupe 2012", class 106 "Ford F-450 Super Duty Crew Cab", and class 134 "Hyundai Sonata Hybrid 2012". This research aims to evaluate the results of vehicle type classification using the Haar wavelet transformation algorithm as a feature extraction method. The feature extraction results of the first and second levels of the Haar wavelet transformation algorithm were subsequently used as training data for the classification using Support Vector Machine (SVM) and K-Nearest Neighbor (KNN). The highest classification result on the first level of training data was in Weighted KNN and Medium SVM with 0.33%. Moreover, the highest classification of the second level of training data was in Weighted KNN with 0.66%. The highest classification result of the combination training data between the first and second levels was in Weighted KNN with 0.75%. [ABSTRACT FROM AUTHOR]

The spine is one part of the human axial skeleton that serves as the body's primary support. Hence, the health of the spine must be considered. The most common spinal abnormality is scoliosis, with the shape of the spine forming the C and S letters. Along with technology development, spinal abnormalities can be identified using images from X-rays to be processed digitally to help health experts as a second opinion to carry out diagnostics of spinal disorders efficiently and accurately. This research was conducted by designing an image processing system for two spine types, normal and abnormal (i.e., scoliosis), by applying the Gray Level Co-occurrence Matrix (GLCM) feature extraction method and two classification methods: K-Nearest Neighbors (KNN) and Support Vector Machine (SVM). The design of this system aims to determine how effective the method is to classify the spine accuracy. The system accuracy in the KNN method reached 73% at a pixel distance of 100 and a quantization level of 16. For the SVM method, the system accuracy value of 90% was obtained at a pixel distance of 75 and a quantization level of 8. The SVM results achieved better than the KNN. [ABSTRACT FROM AUTHOR]

An enormous number of children die due to pneumonia every year worldwide. Pneumonia is a large-scale cause of death amongst children, with a high prevalence rate in South Asia and Sub-Saharan Africa. Even in a developed country like the United States, pneumonia is among the top 10 causes of death. Early detection and treatment of pneumonia can significantly scale down fatality rates among children in countries with a high prevalence. Hence, this paper presents a Convolutional Neural Network model to detect pneumonia using x-ray images. A ResNet50 and VGG-16 pre-trained model was trained to classify x-ray images into two classes, viz., pneumonia and non-pneumonia, by changing various parameters, hyperparameters, and the number of convolutional layers. The VGG-16 model has better performance than ResNet50. Meanwhile, ResNet50 achieved the testing time faster than VGG-16. The CNN model had a better performance in classifying pneumonia and non-pneumonia images. [ABSTRACT FROM AUTHOR]

Tremor is a condition where the patient's body part moves involuntarily in an uncontrollable rhythmic motion. The current treatment includes using medicine, brain surgery or wearable devices. The wearable devices are found to induce lower risk but the results are encouraging. Nevertheless, wearable devices are weighty and bulky. In this study, a novel essential tremor suppression device was developed. This paper aims to perform a feasibility study on measuring angular velocity by validating the mechanical vibration, represented as tremor suppression. As a pilot study, the tremor condition on actual patient was imitated utilising simulator device with a DC motor attached to a small propeller. A10g weight was attached to the propeller's end to produce an unbalanced rotating motion towards the wearer. This additional mass will cause the system rotational inertia, imitating the tremor. For the suppression mechanism, a few vibration motors were attached as actuators. Arduino UNO was used as the microcontroller for both simulator and suppression control systems. The suppression of the tremor was tested by varying the vibration motor arrangements. Finally, vibration analyses were performed at three-axis angular velocity to compare and evaluate the simulated and suppressed tremors. The results confirm that the simulated tremors are in good agreement with literature; and the suppression system successfully reduced the simulated tremor. [ABSTRACT FROM AUTHOR]

Toddlers need an excellent nutritional intake to grow up healthy and get used to a healthy lifestyle. Determining nutritional status is also equally important because it can monitor proper health growth in toddlers. Dietary disorders, including malnutrition, are caused by insufficient food levels and the amount of food not for the needs of each person's body. There is a special health service to monitor the health of toddlers and their vaccinations called an integrated service center. Examination and assessment of the nutritional status of toddlers are carried out in public health services such as an integrated service center. The report of toddler data is still recorded non-digitally, which causes much time to classify their nutrition and data stack. Hence, the nutritional predicate of toddlers can not be monitored constantly. For this reason, a system was built to classify and monitor the nutritional status of toddlers to avoid data accumulation. This system organized the food of toddlers based on anthropometric age weight. Classifying nutrition toddlers digitally using data mining with the C4.5 algorithm included data collection, data selection, data transformation, data mining, and ended with evaluation and knowledge presentation. The results of this study digitally classified toddlers' nutrition using data mining with the C4.5 algorithm at the integrated service centers in Ciasihan Village, with an evaluation value rate of 92% using 50 testing data. In conclusion, the system could classify toddlers' nutrition with data mining using the C4.5 algorithm and help the RED integrated service center officers in the Ciasihan Village determine the nutritional classification of toddlers appropriately. [ABSTRACT FROM AUTHOR]