【摘要】 生物组织变性识别是监测HIFU治疗过程的一个重要方面,对提升HIFU治疗效果有重要意义。提出了一种基于核主元分析(KPCA)和模糊支持向量机(FSVM)的猪肉组织变性分类识别方法。对HIFU辐照离体猪肉组织产生的超声回波信号分别提取能量、衰减系数、背向散射积分等时频域特征,构造表征组织变性特征的混合域特征集。利用KPCA对特征集中能敏感地体现组织变性的特征进行二次特征提取,按累计贡献率高于95%的标准,选择前2个核主元当作主要特征并将其联合输入FSVM进行组织变性识别。实验结果显示,联合特征比单个特征能更好地检测组织变性状态,更准确地对猪肉组织进行变性分类识别。该方法可为监测HIFU治疗中生物组织是否变性提供参考。更多还原

【Abstract】 Identification of biological tissue denaturation is an important aspect for HIFU monitoring during treatment process, and it is of great significance to improve HIFU treatment effect. A method for classification and identification of pork tissues based on kernel principal component analysis( KPCA) and fuzzy support vector machine(FSVM) was proposed. The time-frequency domain characteristics such as energy, attenuation coefficient, back scattering integral were extracted respectively from ultrasonic echo signals generated by HIFU irradiation in vitro pork tissue, and the hybrid domain features set was constructed by above features of tissue degeneration. The features which can sensitively reflect tissue degeneration were secondary extracted by KPCA.According to the criterion of 95% cumulative contribution rate, the first two nuclear elements were selected as the main features, and then added to the FSVM for tissue denaturation recognition. The experimental results show that the united features can detect the organization degeneration state better than the single feature, and can classify the degeneration of pork tissue more accurately. This method can provide a reference for monitoring the degeneration of biological tissues during HIFU therapy.更多还原