集成支持向量机在OCT血管内斑块分割中的应用与研究

OCT医疗图像作为近年来新兴的血管内斑块诊断技术正在迅速成长。为进行有效的OCT血管内斑块辅助语义分割,提出一种基于集成支持向量机(AdaBoost-SVM)的分割学习模型。采用遗传算法GA(Genetic Algorithm)对支持向量机模型的组合参数的选取进行优化以提升其分类性能。通过建立由多个基分类器组合的AdaBoost-SVM集成化分割模型,对OCT血管图像以像素列预测为基础进行准确分割。实验结果表明,在分割过程中相比较BP神经网络以及标准SVM方法,集成支持向量机在以像素列为基础的图像分割过程中具有更优的分割精度。

APPLICATION AND RESEARCH OF ADABOOST-SVM IN OCT INTRAVASCULAR PATCH SEGMENTATION

As an emerging intravascular plaque diagnostic technique, OCT medical image is rapidly growing in recent years. In order to carry out effective OCT intravascular plaque assisted semantic segmentation, we proposed a segmentation learning model based on AdaBoost-SVM. The genetic algorithm( GA) was used to optimize the selection of the combined parameters of the SVM model to improve its classification performance. By establishing an AdaBoost-SVM integrated segmentation model with multiple base classifiers, OCT blood images were segmented accurately on the basis of pixel column prediction. The experimental results show that the Adaboost-SVM has better segmentation accuracy in pixel segmentation-based image segmentation process compared with BP neural network and standard SVM method.