乳腺肿瘤同伦 LM 算法 EIT 图像重建
Image Reconstruction of Breast Tumors Using Homotopy LM Algorithm in Electrical Impedance Tomography

将同伦延拓方法与 Levenberg-Marquardt(LM)迭代算法相结合，对不同的乳腺肿瘤模型进行图像重建，并 与 LM 算法进行比较．在不同迭代初始值下，同伦 LM 算法与 LM 算法的平均电导率相对误差相比，最优值能降低 接近 7%,，运行时间最大能减少 46%,的计算开销，电压相对误差要优于 LM 算法 7%,以上．实验结果表明，同伦 LM 算法比 LM 算法能重建更好的图像，提高了重建图像的精度和收敛速度．
A homotopy Levenberg-Marquardt(HLM)algorithm，which is the combination of homotopy continuation method and Levenberg-Marquardt(LM)algorithm，was presented．The reconstructed images of different breast tumor models were acquired by HLM algorithm and compared with those by LM algorithm．The relative errors of the average conductivity were obtained by means of HLM algorithm and LM algorithm．The optimal value of the relative errors of the average conductivity by HLM algorithm was reduced by nearly 7 percent for different initial values．The best running time and the relative errors of the voltage are decreased by 46 percent and 7 percent respectively．The results show that the better images are reconstructed by HLM algorithm，and the accuracy and convergence rate are improve