基于模式识别的人体肺癌电阻抗检测方法研究

Impedance Detection of Human Lung Cancer Based on Pattern Recognition

根据阻抗频谱理论及模式识别理论,对手术中切除的恶性肿瘤组织及非肿瘤组织进行阻抗测量并进行肺癌辨识.利用Agilent4294A阻抗分析仪测量手术中切除样本的阻抗值,获得100,Hz^100,MHz频率激励下的阻抗频谱,经最小二乘算法拟合,得到频谱特征参数.将测量样本分成训练集与测试集,设计LMSE及Fisher两种线性分类器;分类器经训练后获得判别函数,利用分类器对测试集进行分类实验.研究结果表明,LMSE以及Fisher算法对测试样本均能进行有效识别,并且识别结果与病理切片分析结果吻合,验证了基于模式识别的方法对肺癌阻抗检测进行辨识的可行性与可靠性,为肿瘤早期筛查提供有效检测方法.

According to the impedance spectrum theory and pattern recognition theory,the impedance of the malignant tissue and the non-tumor tissue were measured for identification of lung cancer. With the Agilent4294A impedance analyzer measuring the impedances,the impedance values were obtained under 100,Hz-100,MHz excitation currents. Fitted by the least squares,the parameters in Cole-Cole equation were calculated. All the measured data were divided into a training set and a testing set. Two different linear classifiers based on LMSE and Fisher algorithm were designed. After training,the clas-sification functions were obtained to distinguish the testing set. The research results indicated that LMSE and Fisher algo-rithms can classify the malignant tissue and non-tumor tissue effectively,which is also consistent with the pathological re-sults. It is verified that the method is feasible and reliable in identifying the nature of the tissue,and it can provide an effec-tive screening way for early cancer detection.