目的建立识别中耳内边界的方法。方法利用CT获取耳部结构并建立数值模型;利用Matlab建立神经网络识别程序。结果利用数值模型在鼓膜外侧施加105dB声压,进行谐响应分析获取训练样本;利用训练成熟的网络识别了中耳内边界参数。结论数值识别结果表明,所建的中耳内边界参数识别方法具有良好的识别精度和较快的收敛速度。 Objective To establish a method of identifying the inner ear border. Methods Ear structure was obtained by CT and numerical model was established. Neural network identification program was established by Matlab. Results The numerical model was used to apply the sound pressure of 105 dB to the outer side of the tympanic membrane to obtain the training samples by harmonic response analysis. The middle ear inner boundary parameters were identified by trained networks. Conclusion The results of numerical identification show that the identification method of middle ear parameters in the middle ear has good recognition accuracy and fast convergence rate.