Flexible Strain Sensor Combined with B-ultrasonic Application for Abdominal Organs Deformation Estimation

In recent years, with the rapid development of flexible sensors, various implantable sensors have been developed, and implantable strain sensors also have attracted more attention. However, the development of implantable sensing field still faces many challenges, such as mechanical property matching, biocompatibility, and secondary surgical removal. In this paper, we propose an implantable wireless passive ultrasonic strain sensing system combined with a convolutional neural network (CNN) algorithm to predict the strain rate of the sensor in the proposed human tissue by CNN and then estimate the deformation of the target. The sensing system was a hydrogel compositing with nanohydroxyapatite (nHA), which had good stretchability, biocompatibility, and B-ultrasound visibility in vivo. Meanwhile, as its linear trend property of cross-section with strain, the ultrasound cross-sectional images of the sensor under different strains are extracted using CNN and divided into training and test sets according to the ratio of 8:2 for model training and evaluation. Finally, AlexNet has the best prediction result (91.70% accuracy) among the four selected CNN classification algorithms. It shows that the proposed scheme is an effective choice for predicting organ deformation.