Article
Geography, Physical
K. R. Akshatha, A. K. Karunakar, B. Satish Shenoy, K. Phani Pavan, Chinmay V. Dhareshwar, Dennis George Johnson
Summary: Intelligent UAV video analysis has gained attention for its potential in computer vision applications. In order to address the challenge of small object detection, a Manipal-UAV person detection dataset was created, consisting of images captured from UAVs in varying conditions. The dataset provides a benchmark for evaluating state-of-the-art object detection algorithms on small person objects in aerial view scenarios. The dataset is publicly available for researchers to advance UAV and small object detection research.
ISPRS JOURNAL OF PHOTOGRAMMETRY AND REMOTE SENSING
(2023)
Article
Computer Science, Information Systems
S. Jasmine Minija, M. Anline Rejula, B. Shamina Ross
Summary: Diabetes is a common disease that affects various parts of the body. Diabetic Retinopathy can cause Diabetes Mellitus and potentially lead to blindness. Detecting and classifying Diabetic Retinopathy at an early stage is crucial in preventing blindness.
MULTIMEDIA TOOLS AND APPLICATIONS
(2023)
Article
Chemistry, Analytical
Cheena Mohanty, Sakuntala Mahapatra, Biswaranjan Acharya, Fotis Kokkoras, Vassilis C. Gerogiannis, Ioannis Karamitsos, Andreas Kanavos
Summary: Diabetic retinopathy (DR) is a common complication of long-term diabetes that can cause permanent blindness. This study proposes two deep learning architectures, a hybrid network and DenseNet 121 network, for the detection and classification of DR. The DenseNet 121 model achieved superior performance and has potential for early detection and classification of DR.
Article
Computer Science, Interdisciplinary Applications
Victor Vives-Boix, Daniel Ruiz-Fernandez
Summary: This study presents an automated method for the early detection of diabetic retinopathy using convolutional neural networks with a bio-inspired approach based on synaptic metaplasticity. Results show that including synaptic metaplasticity in the convolutional neural network architectures improves learning rate and accuracy. The best performance was achieved with the InceptionV3 architecture with synaptic metaplasticity.
COMPUTER METHODS AND PROGRAMS IN BIOMEDICINE
(2021)
Article
Medicine, General & Internal
Marc Baget-Bernaldiz, Romero-Aroca Pedro, Esther Santos-Blanco, Raul Navarro-Gil, Aida Valls, Antonio Moreno, Hatem A. Rashwan, Domenec Puig
Summary: The study aimed to test a deep learning algorithm (DLA) in reading retinographies. Results showed that the DLA performed well in identifying any DR and referable DR, with high accuracy, sensitivity, and specificity rates.
Article
Multidisciplinary Sciences
Sajib Saha, Janardhan Vignarajan, Shaun Frost
Summary: This paper presents a computationally efficient and memory efficient CNN-based system for automated detection of glaucoma. The system achieves high accuracy and speed while minimizing resource requirements. It performs well in classifying glaucomatous and non-glaucomatous images, making it suitable for integration into portable fundus cameras.
SCIENTIFIC REPORTS
(2023)
Article
Chemistry, Multidisciplinary
A. M. Mutawa, Shahad Alnajdi, Sai Sruthi
Summary: The serious complication of diabetes, diabetic retinopathy (DR), can lead to vision loss. This study addresses the challenge of varying retinal features in different datasets and proposes a method to efficiently learn and classify DR. By combining multiple datasets, the model's performance improves, allowing healthcare workers to refer patients to ophthalmologists before DR becomes serious.
APPLIED SCIENCES-BASEL
(2023)
Article
Engineering, Biomedical
Rajasekhar Kommaraju, M. S. Anbarasi
Summary: Diabetic Retinopathy (DR) is a serious retinal disease that requires early detection to prevent blindness. Researchers proposed a deep learning model called DRCNNRB to detect blood abnormalities in retinal images. This model utilizes data augmentation and image preprocessing techniques to address the problem of data imbalance and achieve better performance compared to other approaches.
BIOMEDICAL SIGNAL PROCESSING AND CONTROL
(2024)
Article
Biology
Haomiao Liu, Lu Teng, Linhua Fan, Yabin Sun, Huiying Li
Summary: Diabetic retinopathy (DR) is a common early complication of diabetes and a major cause of blindness. To improve the accuracy of DR grading, a dataset of 101 ultra-wide-field (UWF) DR fundus images was established, and a deep learning automatic classification method based on a new preprocessing method was proposed. Experimental results showed high classification accuracy using only the backbone model, with the best-performing Swin-S model achieving ACA 0.72, Macro F1 0.7018, and Kappa 0.65. DR grading using UWF images can achieve higher accuracy and efficiency, with practical significance and value in clinical applications.
COMPUTERS IN BIOLOGY AND MEDICINE
(2023)
Article
Computer Science, Artificial Intelligence
Maryam Dinpajhouh, Seyyed Ali Seyyedsalehi
Summary: We propose an Attention-DenseNet model for detecting and grading the severity of diabetic retinopathy (DR). The model uses a pre-trained convolutional neural network to extract features and hierarchical representation of color fundus images. By using an attention model, the model focuses more on distinctive areas and achieves superior performance in DR detection.
NEURAL COMPUTING & APPLICATIONS
(2023)
Article
Chemistry, Multidisciplinary
Dimple Nagpal, Najah Alsubaie, Ben Othman Soufiene, Mohammed S. Alqahtani, Mohamed Abbas, Hussain M. Almohiy
Summary: Diabetic retinopathy is a complication of diabetes that damages the blood vessels in the eyes, leading to the need for early detection. Transfer learning, using pre-trained models, has been successfully applied to classify diabetic retinopathy.
APPLIED SCIENCES-BASEL
(2023)
Article
Computer Science, Artificial Intelligence
Xiaoling Luo, Zuhui Pu, Yong Xu, Wai Keung Wong, Jingyong Su, Xiaoyan Dou, Baikang Ye, Jiying Hu, Lisha Mou
Summary: The novel diabetic retinopathy detection convolutional network integrates multi-view fundus images to improve the accuracy of DR grading. By utilizing lesions features from the retina within a field-of-view of 120 to 150 degrees, introducing attention mechanisms, and assigning large weights to important channels, the proposed method outperforms other benchmarking methods. The experimental results on a multi-view DR dataset show the effectiveness and superiority of the proposed method for automatic DR detection.
PATTERN RECOGNITION
(2021)
Article
Computer Science, Artificial Intelligence
Zeynab Barzegar, Mansour Jamzad
Summary: Automatic brain abnormality detection is a major challenge in medical image processing. This study presents an efficient and automated algorithm for brain tumour segmentation based on the symmetry of brain structures. The algorithm uses symmetry to determine lesion differences and extract cancerous regions, improving speed and accuracy. Experimental results show performance indices close to manual lesion demarcation accuracy.
IET COMPUTER VISION
(2021)
Article
Chemistry, Multidisciplinary
Atsushi Teramoto, Tomoyuki Shibata, Hyuga Yamada, Yoshiki Hirooka, Kuniaki Saito, Hiroshi Fujita
Summary: By utilizing the novel U-Net R-CNN object detection model along with DenseNet169 convolutional neural network for box classification, the efficiency and accuracy of automated detection of early gastric cancer from endoscopic images can be improved.
APPLIED SCIENCES-BASEL
(2021)
Article
Chemistry, Analytical
Mahmoud Elgafi, Ahmed Sharafeldeen, Ahmed Elnakib, Ahmed Elgarayhi, Norah S. Alghamdi, Mohammed Sallah, Ayman El-Baz
Summary: This study presents a three-step system for detecting diabetic retinopathy using optical coherence tomography (OCT). Experimental results show that the proposed system outperforms related methods and achieves a high accuracy rate.
Article
Computer Science, Artificial Intelligence
Shui-Hua Wang, Suresh Chandra Satapathy, Man-Xia Xie, Yu-Dong Zhang
Summary: COVID-19 is a single-stranded RNA virus, caused by the SARS-CoV-2 strain of coronavirus. This study proposes an ELU-based CNN model for COVID-19 diagnosis, achieving a sensitivity of 94.41 +/- 0.98, specificity of 94.84 +/- 1.21, accuracy of 94.62 +/- 0.96, and F1 score of 94.61 +/- 0.95. The ELUCNN model and mobile app are effective and outperform 14 state-of-the-art COVID-19 diagnosis models in terms of accuracy.
Article
Computer Science, Information Systems
Xiang Yu, Shui-Hua Wang, Yu-Dong Zhang
Summary: To facilitate faster breast cancer detection, a novel and efficient patch-based breast mass detection system was developed. The system consists of three modules: pre-processing, multiple-level breast tissue segmentation, and final breast mass detection. An improved Deeplabv3+ model is used for pre-processing and a multiple-level thresholding segmentation method is proposed for breast mass segmentation. Deep learning models are trained to classify image patches into breast mass or background, reducing the false positive rate. The proposed method achieves comparable performance with state-of-the-art methods.
JOURNAL OF KING SAUD UNIVERSITY-COMPUTER AND INFORMATION SCIENCES
(2023)
Review
Imaging Science & Photographic Technology
Xue Han, Zuojin Hu, Shuihua Wang, Yudong Zhang
Summary: According to the World Health Organization, there have been 625,248,843 confirmed cases of COVID-19, including 65,622,281 deaths worldwide as of October 25, 2022. The diagnosis of COVID-19 using chest X-ray or CT images based on convolutional neural networks (CNN) is an important method for reducing misdiagnosis. This paper introduces the latest deep learning methods and techniques for COVID-19 diagnosis and analyzes existing CNN automatic diagnosis systems, concluding that CNN has essential value in COVID-19 diagnosis and can be further improved with expanded datasets and advanced techniques.
JOURNAL OF IMAGING
(2023)
Article
Computer Science, Artificial Intelligence
Xujing Yao, Ziquan Zhu, Cheng Kang, Shui-Hua Wang, Juan Manuel Gorriz, Yu-Dong Zhang
Summary: This article introduces a computer-aided diagnosis system based on deep learning that automatically classifies chest CT scans into COVID-19, tuberculosis, and healthy control subjects. The system uses a novel classification model called AdaD-FNN, which sequentially transfers trained knowledge and updates sample weights to improve learning of complex patterns. Additionally, a novel image preprocessing model called F-U2MNet-C is used to enhance image features and eliminate interference factors. Experimental results show that the system achieves high classification accuracies for COVID-19 detection and outperforms 22 state-of-the-art methods.
IEEE TRANSACTIONS ON EMERGING TOPICS IN COMPUTATIONAL INTELLIGENCE
(2023)
Article
Computer Science, Artificial Intelligence
Xing Guo, Siyuan Lu, Shuihua Wang, Zhihai Lu, Yudong Zhang
Summary: This paper proposes a facial expression recognition method based on a double-code LBP-layer spatial-attention network (DLSANet) to improve the accuracy of FER. The DLSANet achieves recognition accuracies of 93.81% and 98.68% on the JAFFE and CK+ datasets, respectively, outperforming state-of-the-art methods.
IET IMAGE PROCESSING
(2023)
Article
Engineering, Electrical & Electronic
Wenjing Zhou, Mingwei Shen, Min Xu, Guodong Han, Yudong Zhang
Summary: In this paper, a new sparsity-optimised Farrow structure variable fractional delay (SFS-VFD) filter is proposed to address the aperture effect in wideband array. The method reduces the non-zero coefficients by exploiting coefficient (anti-)symmetry and optimising the number and orders of sub-filters. The proposed method achieves stable and fast convergence by solving the formulated cost function with regularisation constraints using the modified three-block alternating direction multiplier method (MTB-ADMM) with core variable correction items.
IET SIGNAL PROCESSING
(2023)
Article
Computer Science, Information Systems
Shuwen Chen, Siyuan Lu, Shuihua Wang, Yiyang Ni, Yudong Zhang
Summary: Blood cells play a vital role in human metabolism and their status can be used for clinical diagnoses. Manual analysis of blood cell classification is time-consuming, but recent advancements in computer vision can free doctors from this tedious task. This paper proposes a novel automated blood cell classification model, SW-ViT, which outperforms state-of-the-art methods in terms of classification accuracy. The proposed SW-ViT can be applied in daily clinical diagnosis.
Editorial Material
Automation & Control Systems
Jerry Chun-Wei Lin, Gautam Srivastava, Yudong Zhang
ASIAN JOURNAL OF CONTROL
(2023)
Article
Computer Science, Artificial Intelligence
Yudong Zhang, Lijia Deng, Hengde Zhu, Wei Wang, Zeyu Ren, Qinghua Zhou, Siyuan Lu, Shiting Sun, Ziquan Zhu, Juan Manuel Gorriz, Shuihua Wang
Summary: Integrating artificial intelligence with food category recognition has been a field of interest for research, and it has the potential to revolutionize human interaction with food. The advancements in big data and deep learning have provided better recognition methods. This survey focuses on machine learning systems for food category recognition, including datasets, data augmentation, feature extraction, and algorithms, with a special emphasis on deep learning techniques.
INFORMATION FUSION
(2023)
Review
Computer Science, Information Systems
Xing Guo, Yudong Zhang, Siyuan Lu, Zhihai Lu
Summary: This study summarizes the methods and datasets of facial expression recognition, including machine learning and deep learning methods, and compares their advantages and limitations. It also concludes the current problems and future development of facial expression recognition.
MULTIMEDIA TOOLS AND APPLICATIONS
(2023)
Review
Computer Science, Artificial Intelligence
Chenxi Huang, Jiaji Wang, Shuihua Wang, Yudong Zhang
Summary: Oral diseases have significant impact on human health, but they are often unnoticed in early stages. Deep learning, as a promising field in artificial intelligence, has achieved remarkable success in various domains, particularly in dentistry. This paper aims to provide an overview of recent research on deep learning applications in dentistry, with a focus on dental imaging. Deep learning algorithms excel in difficult tasks such as image segmentation and recognition, enabling accurate identification of oral conditions and abnormalities. Integration of deep learning with other oral health data offers a holistic understanding of the relationship between oral and systemic health. However, there are still many challenges that need to be addressed.
Editorial Material
Imaging Science & Photographic Technology
Yudong Zhang, Jiaji Wang, Juan Manuel Gorriz, Shuihua Wang
JOURNAL OF IMAGING
(2023)
Article
Biology
Wei Wang, Yanrong Pei, Shui-Hua Wang, Juan Manuel Gorrz, Yu-Dong Zhang
Summary: Since 2019, the COVID-19 pandemic has posed a significant threat to the global economy and human health. Deep learning-based computer-aided diagnosis models can effectively alleviate the challenges of diagnosing COVID-19 due to limited healthcare resources. To overcome the time-consuming and unstable nature of traditional hyperparameter tuning methods, we propose a Particle Swarm Optimization-guided Self-Tuning Convolution Neural Network (PSTCNN) that automatically adjusts the model's hyperparameters.
Article
Oncology
Ziquan Zhu, Shui-Hua Wang, Yu-Dong Zhang
Summary: An automated network model for blood cell classification is proposed, which can assist doctors in diagnosing disease types and severity. The model uses a ResNet50 backbone for feature extraction and applies an ensemble of three randomized neural networks based on majority voting. Experimental results show that the proposed method outperforms other state-of-the-art methods in terms of classification performance.
TECHNOLOGY IN CANCER RESEARCH & TREATMENT
(2023)
Review
Engineering, Multidisciplinary
Xiaoyan Jiang, Zuojin Hu, Shuihua Wang, Yudong Zhang
Summary: This paper introduces the latest methods and various techniques and algorithms of posture recognition, analyzes the general process and datasets, and compares several improved CNN methods and three main recognition techniques. Additionally, it discusses the applications of advanced neural networks in posture recognition and highlights the need for further research in feature extraction, information fusion, and data generation.
CMES-COMPUTER MODELING IN ENGINEERING & SCIENCES
(2023)