Biomechanical Modeling of Tissues at Relaxation Stage Based on Neural Network / 医用生物力学

ABSTRACT

Objective Taking pig kidney as an example, through a series of comparative and analogical experiments, the influencing factors of compressive stress at relaxation stage of biological tissues were analyzed, and a more accurate and widely applicable biomechanical model at relaxation stage was established. Methods The compressive stress relaxation experiments of pig kidney under different conditions were carried out by using the self-built mechanical experiment platform. The collected data were analyzed and mapped, and various factors affecting the relaxation force changes were summarized. Based on the conclusion, the neural network learning algorithm was used to model the force change process at relaxation stage of pig kidney. Results The pre-extrusion pressure and relaxation time were the main influencing factors for compressive stress changes of biological tissues at relaxation stage. The average error of test sample validation experiment was 6.4 mN, and the average prediction error of generalization sample validation experiment was 34.9 mN, so the modeling effect was good. Conclusions Neural network modeling algorithm has the advantages of strong generalization ability and good fault tolerance, which contributes to providing more realistic force tactile feedback prediction for virtual surgery system. It is also a new idea for mechanical modeling of nonlinear biological tissues.