Article
Computer Science, Artificial Intelligence
S. Qasim Abbas, Lianhua Chi, Yi-Ping Phoebe Chen
Summary: Structural magnetic resonance imaging (sMRI) is a prevalent and potent imaging modality for the computer-aided diagnosis (CAD) of neurological diseases like dementia. Recent advancements in deep learning, particularly convolutional neural networks (CNNs), have shown promise in diagnosing Alzheimer's disease (AD) by learning the atrophy patterns in sMRIs. However, the current CNN-based approaches still need to improve their diagnostic performance. To address this issue, the proposed three-dimensional Jacobian domain convolutional neural network (JD-CNN) offers excellent classification performance without the need for landmark detection. The JD-CNN model is trained based on features transformed from the spatial domain to the Jacobian domain, and it surpasses previously reported state-of-the-art techniques in terms of classification performance.
PATTERN RECOGNITION
(2023)
Article
Computer Science, Artificial Intelligence
Brendan Kolisnik, Isaac Hogan, Farhana Zulkernine
Summary: We propose a hierarchical image classification model, Condition-CNN, which improves prediction accuracy and reduces training time by using the Teacher Forcing training algorithm and conditional probabilities. The validation results show that Condition-CNN achieves higher prediction accuracy for Level 1, 2, and 3 classes compared to other baseline CNN models.
EXPERT SYSTEMS WITH APPLICATIONS
(2021)
Article
Computer Science, Interdisciplinary Applications
Zhentao Hu, Zheng Wang, Yong Jin, Wei Hou
Summary: This study proposes a VGG-TSwinformer model based on convolutional neural network (CNN) and Transformer for short-term longitudinal study of MCI. This model utilizes VGG-16 based CNN to extract low-level spatial features of longitudinal sMRI images and maps them to high-level feature representations. Sliding-window attention is used for fine-grained fusion of spatially adjacent feature representations, and temporal attention is used to measure the evolution of these feature representations. The model achieved high accuracy, sensitivity, specificity, and AUC in the classification task of sMCI vs pMCI on the ADNI dataset.
COMPUTER METHODS AND PROGRAMS IN BIOMEDICINE
(2023)
Article
Computer Science, Artificial Intelligence
Serhat Kilicarslan, Cemil Kozkurt, Selcuk Bas, Abdullah Elen
Summary: Pneumonia is a major global health problem, and deep learning techniques are proposed to diagnose it. This study suggests a method for detecting pneumonia using a new activation function, and experimental results show that CNN models with this activation function achieve the best performance in both pneumonia detection and traditional benchmark datasets.
EXPERT SYSTEMS WITH APPLICATIONS
(2023)
Article
Computer Science, Artificial Intelligence
Ruoyu Yang, Shubhendu Kumar Singh, Mostafa Tavakkoli, M. Amin Karami, Rahul Rai
Summary: Structural health monitoring (SHM) is the implementation of a damage detection strategy for structures. Current methods, like physically attached sensors or non-contact vision-based vibration measurements, have significant drawbacks. Recently, deep learning methods like CNN and FCN have been applied for defect detection, addressing these problems and achieving high accuracy.
APPLIED INTELLIGENCE
(2023)
Article
Computer Science, Artificial Intelligence
Shakila Shojaei, Mohammad Saniee Abadeh, Zahra Momeni
Summary: Deep neural networks (DNN) are widely used in medical tasks, and Convolutional Neural Networks (CNN) have shown excellent performance for image-based tasks. However, CNN lacks the ability to explain its outputs. In this study, a 3D-CNN model was trained using MRI scans to diagnose Alzheimer's Disease (AD) patients. By combining a genetic algorithm-based Occlusion Map method with a set of Backpropagation-based explainability methods, the model achieved an accuracy of 87% in 5-fold cross-validation, demonstrating its effectiveness.
EXPERT SYSTEMS WITH APPLICATIONS
(2023)
Article
Computer Science, Artificial Intelligence
Qiang Liu, Xiangchao Meng, Feng Shao, Shutao Li
Summary: This paper proposes a Supervised-Unsupervised combined Fusion Network (SUFNet) for high-fidelity pansharpening. It integrates multiscale mechanisms, dilated convolution, and skip connection to improve the robustness of the network, and uses an Unsupervised Spatial-Spectral Compensation Network (USSCNet) to enhance the spatial and spectral fidelity. The experiments demonstrate the competitive performance of the proposed method.
INFORMATION FUSION
(2023)
Article
Astronomy & Astrophysics
Kai Feng, Long Xu, Dong Zhao, Sixuan Liu, Xin Huang
Summary: Timely solar flare forecasting is hindered by data transmission delay, prompting the need to deploy compression models on satellites. Three compression methods (knowledge distillation, pruning, and quantization) were examined, along with a proposed assembled compression model that demonstrated effective compression and maintained accuracy.
ASTROPHYSICAL JOURNAL SUPPLEMENT SERIES
(2023)
Article
Computer Science, Artificial Intelligence
Hua Li, Jing Niu, Dengao Li, Chen Zhang
Summary: This study utilizes deep learning to automatically classify breast masses in mammograms into benign and malignant. By combining convolutional neural network (CNN) and recurrent neural network (RNN), the model proposed in the study significantly improves the performance of benign and malignant breast mass classification.
IET IMAGE PROCESSING
(2021)
Article
Biology
Fei Liu, Huabin Wang, Shiuan-Ni Liang, Zhe Jin, Shicheng Wei, Xuejun Li
Summary: Structural magnetic resonance imaging (sMRI) is widely used in the diagnosis of Alzheimer's disease (AD), but only a few specific atrophy areas in sMRI scans are strongly associated with AD. The challenge lies in identifying discriminating atrophy features between patients, as the degree of atrophy and lesion areas differ among individuals. To address this, we propose a MPS-FFA model that combines multiplane and multiscale feature-level fusion attention.
COMPUTERS IN BIOLOGY AND MEDICINE
(2023)
Article
Neurosciences
Meera Srikrishna, Joana B. Pereira, Rolf A. Heckemann, Giovanni Volpe, Danielle van Westen, Anna Zettergren, Silke Kern, Lars-Olof Wahlund, Eric Westman, Ingmar Skoog, Michael Scholl
Summary: In this study, an automatic method for segmenting grey matter, white matter, cerebrospinal fluid, and intracranial volume from head CT images using a U-Net deep learning model trained on MRI-derived segmentation labels was proposed and validated. Results showed accurate prediction of brain tissue classes from unseen CT images, demonstrating the potential for CT-derived segmentations to be used in clinical practice and research.
Article
Computer Science, Artificial Intelligence
Arati Paul, Sanghamita Bhoumik, Nabendu Chaki
Summary: A novel deep learning framework utilizing convolutional neural networks is proposed for feature extraction in hyperspectral image classification. Experimental results demonstrate the superiority of the proposed model in effectively classifying hyperspectral images.
NEURAL COMPUTING & APPLICATIONS
(2021)
Article
Mathematics
Yu-Hung Tsai, Sheng-Kuang Wu, Shyr-Shen Yu, Meng-Hsiun Tsai
Summary: The exploration of elite athletes' performance through cognitive neuroscience has recently emerged as a research method. Most studies on cognitive abilities and athletic performance focus on closed skills rather than open skills. This study collected dynamic brain wave data of table tennis athletes during specific plays and used deep neural network algorithms to predict their performance. By converting the data from time domain to frequency domain and using a hybrid convolutional neural network framework, the proposed algorithm achieved an accuracy of 96.70% in predicting athletes' performance. This study contributes to the understanding of dynamic brain waves in open skills and creates a novel classification model for identifying brain waves associated with good elite sports performance.
Article
Biotechnology & Applied Microbiology
Xiyue Wang, Ruijie Wang, Sen Yang, Jun Zhang, Minghui Wang, Dexing Zhong, Jing Zhang, Xiao Han
Summary: This study proposes an innovative two-stage model for glioma classification based on radiology and histology data. The model achieved first place in the MICCAI 2020 CPM-RadPath Challenge and demonstrated high performance on the validation set.
FRONTIERS IN BIOENGINEERING AND BIOTECHNOLOGY
(2022)
Article
Remote Sensing
Wei Zhang, Ping Tang, Lijun Zhao
Summary: Traditional CNN methods for land-cover classification have issues with high computation cost and low efficiency, while methods based on FCN have opened up new possibilities for efficient land-cover classification.
INTERNATIONAL JOURNAL OF REMOTE SENSING
(2021)