Alzheimer’s disease is a neurodegenerative disorder prevalent in older adults, and early diagnosis is crucial for effective treatment. A deep learning model can automatically classify Alzheimer’s disease from magnetic resonance imaging to aid clinicians in diagnosis. Convolutional Neural Networks (CNNs) are commonly used for disease detection in medical images, but their performance is limited due to inadequate labeled data, high inter-class similarity, and overfitting problems. Key hyperparameters influencing CNN performance include the number of convolution layers and filters assigned to each convolution layer. About other hyperparameters, numerous combinations exist. Since CNN models take a long time to train, it is quite costly to try all combinations to find the optimal model. Existing studies have optimized only a few hyperparameters, such as learning rate, batch size, and optimizer in custom and transfer learning models. In this study, we propose an algorithm based on particle swarm optimization to fine-tune the hyperparameters, including the number of convolution layers, filters, and other hyperparameters, in CNN architectures designed to classify Alzheimer’s disease severity. Using the proposed lightweight model, Alzheimer’s disease was accurately classified with an accuracy of 99.53% and an F1-score of 99.63% on a public dataset. Our model surpasses the performance of previous studies, offering the potential to alleviate the burden on doctors and expedite their decision-making processes. The developed framework can be accessed via the link: “https://ai.gop.edu.tr/alzheimer”.
2 May 2024 in Engineering Applications of Artificial Intelligence
Brain tumors can have fatal consequences, affecting many body functions. For this reason, it is essential to detect brain tumor types accurately and at an early stage to start the appropriate treatment process. Although convolutional neural networks (CNNs) are widely used in disease detection from medical images, they face the problem of overfitting in the training phase on limited labeled and insufficiently diverse datasets. The existing studies use transfer learning and ensemble models to overcome these problems. When the existing studies are examined, it is evident that there is a lack of models and weight ratios that will be used with the ensemble technique. With the framework proposed in this study, several CNN models with different architectures are trained with transfer learning and fine-tuning on three brain tumor datasets. A particle swarm optimization-based algorithm determined the optimum weights for combining the five most successful CNN models with the ensemble technique. The results across three datasets are as follows: Dataset 1, 99.35% accuracy and 99.20 F1-score; Dataset 2, 98.77% accuracy and 98.92 F1-score; and Dataset 3, 99.92% accuracy and 99.92 F1-score. We achieved successful performances on three brain tumor datasets, showing that the proposed framework is reliable in classification. As a result, the proposed framework outperforms existing studies, offering clinicians enhanced decision-making support through its high-accuracy classification performance.
31 August 2023 in Sakarya University Journal of Computer and Information Sciences
With urbanization, population, and consumption on the rise, urban waste generation is steadily increasing. Consequently, waste management systems have become integral to city life, playing a critical role in resource efficiency and environmental protection. Inadequate waste management systems can adversely affect the environment, human health, and the economy. Accurate and rapid automatic waste classification poses a significant challenge in recycling. Deep learning models have achieved successful image classification in various fields recently. However, the optimal determination of many hyperparameters is crucial in these models. In this study, we developed a deep learning model that achieves the best classification performance by optimizing the depth, width, and other hyperparameters. Our six-layer Convolutional Neural Network (CNN) model with the lowest depth and width produced a successful result with an accuracy value of 89% and an F1 score of 88%. Moreover, several state-of-the-art CNN models such as VGG19, DenseNet169, ResNet101, Xception, InceptionV3, RegnetX008, RegnetY008, EfficientNetV2S trained with transfer learning and fine-tuning. Extensive experimental work has been done to find the optimal hyperparameters with GridSearch. Our most comprehensive DenseNet169 model, which we trained with fine-tuning, provided an accuracy value of 96.42% and an F1 score of 96%. These models can be successfully used in a variety of waste classification automation.