基于马氏距离分类准则的室间隔缺损诊断研究

On the Mahalanobis Distance Classification Criterion for a Ventricular Septal Defect Diagnostic System

摘要: 本研究提出一种基于马氏距离的分类准则实现对三种室间隔缺损(VSD)的诊断。与诊断方法对应的三个阶段概括如下。第一阶段，通过电子听诊器采集心音，然后利用小波分解进行预处理。在第二阶段提取时频特征并进行主成分分析(PCA)降维。第三阶段，描述了基于马氏距离分类准则的VSD诊断方法。最后，通过与其它诊断VSD的分类方法进行比较来评价本方法效果。分析研究结果，本研究对三种室间隔缺损与正常心音的分类精度分别为95.2%、94.4%、97.1%、99.1%，优于其它知名分类器。因此本研究可为医护人员或患者诊断VSD提供一种有效的方法。

Abstract: This study proposes a criterion based on the Mahalanobis distance for diagnosing three-type ventricular septal defects (VSDs). The three stages corresponding to the diagnostic method are generally summarized as follows. In the first stage, the heart sound is collected via an electronic stethoscope and preprocessed using the wavelet decomposition. The time-frequency features are extracted in the second stage. And finally, the third stage describes the Mahalanobis distance classification criterion-based diagnostic method used to diagnose the VSD. The performance of this proposed method is evaluated by comparing with other well-knows classification methods in diagnosing sounds from patients with VSDs. The classification accuracy of three-kind of VSDs and normal heart sound are 95.2%, 94.4%, 97.1%, and 99.1%, respectively, which are greater than other well-known classifier methods. Therefore, the proposed method can provide an efficient way to diagnose VSD for medical staff or patients.