Bioimpedance spectroscopy for lung cancer diagnosis based on Hippopotamus algorithm with Enhanced exploitation capability (EEC-HA)

Abstract

A Hippopotamus Algorithm with Enhanced Exploitation Capability (EEC-HA) is proposed in this study to improve the diagnostic capability of Bioimpedance Spectroscopy (BIS) for lung cancer. Firstly, an EEC-HA algorithm is developed to fit the equivalent circuit resistance R_P by combining the hippopotamus exploration strategy and multi-dimensional exploitation strategy. Secondly, the Distribution of Relaxation-Time based on Gaussian Discretization (GDRT) is applied to analyze the relaxation peak γ_peak of the tissues. Thirdly, 39 samples each of lung cancer and normal tissues are collected. Then, the significant differences of R_P and γ_peak are statistically analyzed with AUC (Area Under Curve) of 0.863 and 0.814, respectively. Multivariate analysis is performed to establish regression equations $\text{RF = 0.3689}{R}_P\text{+ 0.9295}{\gamma }_{\mathrm{peak}}$ by combining the above two parameters. Finally, pathological tissue experiments verified the performance of EEC-HA. The linear correlation R² of EEC-HA is improved by 3.4% and 2%, with the average fitting error reduced by 69.2% and 46.3% compared with the Whale Optimization Algorithm and Hippopotamus Algorithm, respectively. The AUC of the regression coefficient RF is 0.913, with a cut-off value of 787.04, and the sensitivity and specificity are 0.897 and 0.897, respectively. Therefore, the proposed BIS method combining EEC-HA and GDRT is effective in lung cancer diagnosis, which is expected to be applied to intraoperative detection in the future.