In clinic, intima and media thickness are the main indicators for evaluating the development of atherosclerosis. At present, these indicators are measured by professional doctors manually marking the boundaries of the inner and media on B-mode images, which is complicated, time-consuming and affected by many artificial factors. A grayscale threshold method based on Gaussian Mixture Model (GMM) clustering is therefore proposed to detect the intima and media thickness in carotid arteries from B-mode images in this paper. Firstly, the B-mode images are clustered based on the GMM, and the boundary between the intima and media of the vessel wall is then detected by the gray threshold method, and finally the thickness of the two is measured. Compared with the measurement technique using the gray threshold method directly, the clustering of B-mode images of carotid artery solves the problem of gray boundary blurring of inner and middle membrane, thereby improving the stability and detection accuracy of the gray threshold method. In the clinical trials of 120 healthy carotid arteries, means of 4 manual measurements obtained by two experts are used as reference values. Experimental results show that the normalized root mean square errors (NRMSEs) of the estimated intima and media thickness after GMM clustering were 0.104 7 ± 0.076 2 and 0.097 4 ± 0.068 3, respectively. Compared with the results of the direct gray threshold estimation, means of NRMSEs are reduced by 19.6% and 22.4%, respectively, which indicates that the proposed method has higher measurement accuracy. The standard deviations are reduced by 17.0% and 21.7%, respectively, which indicates that the proposed method has better stability. In summary, this method is helpful for early diagnosis and monitoring of vascular diseases, such as atherosclerosis.
Existing classification methods for myositis ultrasound images have problems of poor classification performance or high computational cost. Motivated by this difficulty, a lightweight neural network based on a soft threshold attention mechanism is proposed to cater for a better IIMs classification. The proposed network was constructed by alternately using depthwise separable convolution (DSC) and conventional convolution (CConv). Moreover, a soft threshold attention mechanism was leveraged to enhance the extraction capabilities of key features. Compared with the current dual-branch feature fusion myositis classification network with the highest classification accuracy, the classification accuracy of the network proposed in this paper increased by 5.9%, reaching 96.1%, and its computational complexity was only 0.25% of the existing method. The obtained results support that the proposed method can provide physicians with more accurate classification results at a lower computational cost, thereby greatly assisting them in their clinical diagnosis.
Early screening is an important means to reduce breast cancer mortality. In order to solve the problem of low breast cancer screening rates caused by limited medical resources in remote and impoverished areas, this paper designs a breast cancer screening system aided with portable ultrasound Clarius. The system automatically segments the tumor area of the B-ultrasound image on the mobile terminal and uses the ultrasound radio frequency data on the cloud server to automatically classify the benign and malignant tumors. Experimental results in this study show that the accuracy of breast tumor segmentation reaches 98%, and the accuracy of benign and malignant classification reaches 82%, and the system is accurate and reliable. The system is easy to set up and operate, which is convenient for patients in remote and poor areas to carry out early breast cancer screening. It is beneficial to objectively diagnose disease, and it is the first time for the domestic breast cancer auxiliary screening system on the mobile terminal.
Ultraviolet radiation is a primary external factor contributing to skin photoaging, as it induces cellular deoxyribonucleic acid damage and collagen degeneration, thereby accelerating skin aging and increasing the risk of skin cancer. Currently, skin aging assessment mainly relies on dermatologists’ empirical judgment, which is inherently subjective and inefficient. To address these limitations, this study proposes a deep learning–driven intelligent skin aging classification method based on high-frequency ultrasound images. The proposed model employs efficient network version 2 (EfficientNetV2) as the backbone network and introduces the Gaussian error linear unit (GELU) activation function and layer normalization to enhance nonlinear feature representation and training stability. In addition, a global-aware temporal hierarchical network is integrated to enable efficient multi-scale feature extraction of skin tissue. A dual enhancement attention mechanism, combining parallel channel–spatial attention and squeeze-and-excitation modules, is further designed to improve the model’s sensitivity to key aging-related regions. Moreover, a multi-scale path dropout regularization strategy is adopted to effectively alleviate overfitting. Experiments conducted on a facial high-frequency ultrasound dataset collected from subjects aged 25~55 years demonstrate that the proposed method achieves an accuracy of 87.66%, a precision of 88.27%, a recall of 87.66%, an F1 score of 87.80%, and a specificity of 97.94%, consistently outperforming existing mainstream models. These results indicate that the proposed approach enables high-precision identification of skin aging levels and provides an efficient and objective auxiliary diagnostic tool for skincare, anti-aging treatment, and the prevention of photoaging-related skin diseases.
Aiming at the limitations of low accuracy and poor stability in the transit time method for estimating carotid local pulse wave velocity, this paper proposed a machine learning-based local pulse wave velocity estimation method, which integrated carotid pulse wave time-domain features, age, and cardiac function parameters. The research was based on a dataset of carotid pulse wave propagation from 4 374 virtual subjects. By combining the Pearson correlation coefficient method and the least absolute shrinkage and selection algorithm to select multi-position combinations of pulse wave time-domain features, and integrating age, heart rate, and other parameters as input features, five machine learning models including multiple linear regression, Bayesian ridge regression, k-nearest neighbor regression, support vector regression and convolutional neural network were used to construct the carotid local pulse wave velocity estimation model, respectively. The results demonstrated that all five machine learning models showed higher accuracy and stronger stability than the traditional methods, and the support vector regression model achieved the optimal performance, with a normalized root mean square error of less than 1.80% and a coefficient of determination exceeding 0.980. In conclusion, it is hoped that the research results presented in this paper can provide a theoretical basis and technical support for the early quantitative assessment of local vascular elasticity of the carotid artery in clinic.