Improve the diagnosis of atrial hypertrophy with the local discriminative support vector machine

Abstract

Abstract.Computer-aided diagnosis (CAD) approaches succeed in detecting a number of diseases, however, they are not good at addressing atrial hypertrophy disease due to the lack of training data. Support Vector Machine (SVM) is very popular in few CAD solutions to atrial hypertrophy. Yet the performance of SVM is moderate in atrial hypertrophy detection compared to its success in other classification problems. In this paper we propose a novel CAD algorithm, Local Discriminative SVM (LDSVM), to overcome the above-mentioned difficulty. LDSVM consists of a global SVM that is trained on the training data, and a local kNN that is trained based on the information of SVM and query. When a query arrives, SVM gives the initial decision. If the initial decision is less confident, local kNN begins to modify the initial decision. LDSVM improves the accuracy of SVM through some contributions: the risk tube, the discriminant information derived from SVM hyperplane, the new metric and the self-tuning size of neighborhood. Empirical evidence on real medical datasets show high performance of LDSVM over the peers in addressing atrial hypertrophy problems.

A Multi-Sensor Data Fusion Approach for Atrial Hypertrophy Disease Diagnosis Based on Characterized Support Vector Hyperspheres

“…In the experiments, for CSVH, class 1 was viewed as the patient class and class 2 was viewed as the health class. Here, besides CSVH, we conducted SVM [ 18 , 19 ]; local discriminative SVM (LDSVM) [ 25 ]; two SVM variants, multi-weight vector projection support vector machines (MVSVM) [ 31 ]; twin support vector machines (TWSVM) [ 32 ]; as well as a neural network, named as NN1. NN1 has three layers, including an input layer, a hidden layer and an output layer.…”

“…Among them, the second file contains electrocardio curves, which were to be investigated. To obtain a fair comparison, we employed the sampling method in [ 25 ] to derive vectors from the electrocardio curves. That is, we chose five heartbeats from the electrocardio curve of one instance to represent the cardiac information.…”

“…SVM computes the support vectors based on which a hyperplane is defined to maximize the separation between two classes. Recently, SVM has played an important role in computer-aided diagnosis for cardiovascular diseases including atrial hypertrophy [ 20 , 21 , 22 , 23 , 24 , 25 ], because it has good generalization performance, a compact structure and a solid theoretical basis. A number of SVM-based works have been proposed in recent years.…”

“…Recently, a diagnosis approach based on sensor data fusion was presented in [ 25 ], a locally discriminant SVM (LDSVM) for atrial hypertrophy diagnosis. As an assembling approach, LDSVM consists of SVM and k-Nearest Neighbors (kNN).…”

“…However, there are some problems encoded with LDSVM. Firstly, as [ 25 ] indicated, the scenarios when the initial SVM decision fails usually involve datasets with overlapping classes. In such cases, kNN works in the overlapping regions based on a metric derived from the SVM hyperplane function.…”