Fully Automatic Analysis of Muscle B-Mode Ultrasound Images Based on the Deep Residual Shrinkage U-Net

The parameters of muscle ultrasound images reflect the function and state muscles. They are great significance to diagnosis diseases. Because manual labeling is time-consuming laborious, automatic image has become a research topic. In recent years, there have been many methods that apply processing deep learning automatically analyze images. However, these limitations, such as being non-automatic, not applicable with complex noise, only able measure single parameter. This paper proposes fully analysis method based on segmentation solve problems. Deep Residual Shrinkage U-Net(RS-Unet) accurately segment Compared existing methods, accuracy our shows improvement. mean differences pennation angle, fascicle length thickness about 0.09°, 0.4 mm 0.63 mm, respectively. Experimental results show proposed realizes accurate measurement exhibits stability robustness.