RESUMEN
Background: Steady state visually evoked potentials (SSVEPs) based early glaucoma diagnosis requires effective data processing (e.g., deep learning) to provide accurate stimulation frequency recognition. Thus, we propose a group depth-wise convolutional neural network (GDNet-EEG), a novel electroencephalography (EEG)-oriented deep learning model tailored to learn regional characteristics and network characteristics of EEG-based brain activity to perform SSVEPs-based stimulation frequency recognition. Method: Group depth-wise convolution is proposed to extract temporal and spectral features from the EEG signal of each brain region and represent regional characteristics as diverse as possible. Furthermore, EEG attention consisting of EEG channel-wise attention and specialized network-wise attention is designed to identify essential brain regions and form significant feature maps as specialized brain functional networks. Two publicly SSVEPs datasets (large-scale benchmark and BETA dataset) and their combined dataset are utilized to validate the classification performance of our model. Results: Based on the input sample with a signal length of 1 s, the GDNet-EEG model achieves the average classification accuracies of 84.11, 85.93, and 93.35% on the benchmark, BETA, and combination datasets, respectively. Compared with the average classification accuracies achieved by comparison baselines, the average classification accuracies of the GDNet-EEG trained on a combination dataset increased from 1.96 to 18.2%. Conclusion: Our approach can be potentially suitable for providing accurate SSVEP stimulation frequency recognition and being used in early glaucoma diagnosis.
RESUMEN
Background: Automated diagnosis of various retinal diseases based on fundus images can serve as an important clinical decision aid for curing vision loss. However, developing such an automated diagnostic solution is challenged by the characteristics of lesion area in 2D fundus images, such as morphology irregularity, imaging angle, and insufficient data. Methods: To overcome those challenges, we propose a novel deep learning model named MyopiaDETR to detect the lesion area of normal myopia (NM), high myopia (HM) and pathological myopia (PM) using 2D fundus images provided by the iChallenge-PM dataset. To solve the challenge of morphology irregularity, we present a novel attentional FPN architecture and generate multi-scale feature maps to a traditional Detection Transformer (DETR) for detecting irregular lesion more accurate. Then, we choose the DETR structure to view the lesion from the perspective of set prediction and capture better global information. Several data augmentation methods are used on the iChallenge-PM dataset to solve the challenge of insufficient data. Results: The experimental results demonstrate that our model achieves excellent localization and classification performance on the iChallenge-PM dataset, reaching AP50 of 86.32%. Conclusion: Our model is effective to detect lesion areas in 2D fundus images. The model not only achieves a significant improvement in capturing small objects, but also a significant improvement in convergence speed during training.
