Search

Diabetic retinopathy is one of the most prevalent microvascular complications of diabetes mellitus, and photoacoustic imaging is an effective method for imaging diabetic retinal vessels. Photoacoustic imaging is an emerging noninvasive imaging method based on the photoacoustic effect, which offers advantages of contrast, resolution, and depth imaging. Appropriate photoacoustic reconstruction methods are essential for obtaining high‐quality photoacoustic images. In this study, a multi‐input self‐attention multiscale feature fusion network (SAMF‐Net) is proposed for photoacoustic reconstruction. The algorithm accepts two inputs, namely the original photoacoustic signal and the traditional reconstructed image. Furthermore, a global feature extraction module based on the self‐attention mechanism is employed to focus on the global information. The results demonstrate that the proposed method exhibits superior reconstruction capability under different sparse detection views. The method has instructive value for photoacoustic image reconstruction and has the potential for further application in the diagnosis of diabetic retinopathy. [ABSTRACT FROM AUTHOR]

Cardiovascular disease (CVD) is a major cause of mortality worldwide. To facilitate early prevention and timely diagnosis of CVD, daily electrocardiogram (ECG) monitoring has become an important tool. However, various noises introduced into ECG signals will affect the diagnostic utility of these recordings. In addition, it is hard for non-specialists to interpret the ECG signals. To address this problem, we propose CS-TRANS, a novel deep learning framework for ECG denoising and classification. This framework mainly incorporates: (i) convolutional neural networks (CNN) regulated by stationary wavelet transform (SWT), called CNN-SWT; and (ii) Transformer encoder. The convolution kernel constraints and architecture of SWT are introduced into CNN to learn both the linear and non-linear time-frequency features more efficiently, while the transformer encoder further boosts global feature extraction. We implemented CS-TRANS and compared it with the state of the arts using the MIT-BIH arrhythmia database. The evaluation results show that compared to other algorithms, CS-TRANS improves the output signal-to-noise ratio (SNR) by 10% in denoising, and achieves 0.5% higher accuracy in classification, while saving half to even more than 90% of the parameters of the model. [ABSTRACT FROM AUTHOR]

The Raman spectroscopy analysis has been applied to the detection and research of oral cancer. One of the essential works in this technique is the Raman spectral data analysis method, which is mainly divided into two categories: traditional machine learning and deep learning. Especially, the deep learning method is proved that it could obtain higher accuracy in oral cancer identification than the traditional machine learning method. The purpose of this study is to test, compare, and analyze the performance of existing classical deep neural network models (AlexNet, VGGNet, ResNet50, MobileNetV2, Transformer) that recognize multiple types of oral cancer tissues. To achieve this goal, 16,200 Raman spectra are first collected from 180 tissue samples of 90 patients who have undergone a surgical resection due to tongue squamous cell carcinoma, gingival squamous cell carcinoma, and buccal squamous cell carcinoma. Then, the models are trained and predicted at the patient level. The experimental results demonstrate that the ResNet50 has the best performance in the identification of oral cancer tissue and normal tissue with an overall accuracy rate of 92.81%, an overall precision rate of 92.93%, and an overall recall rate of 92.86%. With this foundation, we further develop a prototype intelligent detection system with above five classical deep neural network models to realize multi-types of oral cancer tissue detection. Hopefully, our work can provide a guide for oral cancer detection using the deep learning method with Raman spectroscopy analysis and promote the development of clinical diagnosis system for oral cancer. The code is available at https://github.com/ISCLab-Bistu/deep-learning-for-OSCC. [ABSTRACT FROM AUTHOR]