Abstract: Computer Aided Diagnosis（CAD） is already wildly used in many diseases diagnosis.By using CAD，doctor can improve the efficiency of the diagnosis without decreasing the diagnosis accuracy，therefore，it can save much time to help the patients to get more cures.This paper presents a co-classifier for computer aided Pulmonary Embolism diagnosis：At first，an AdaBoost classifier is built by the AdaBoost algorithm with CART（Classification and Regression Trees） as the weak-learners，and then，a new co-classifier is obtained from the combination of the AdaBoost classifier with the BP Artificial Neural Networks.Experiment results show that this co-classifier can improve both the diagnosis efficiency and the diagnosis accurate（evaluated by the detection rate and diagnosis sensitivity）.

Key words: ensemble of learning method, Computer Aided Diagnosis（CAD）, Pulmonary Embolism（PE）, artificial neural networks, AdaBoost algorithm

摘要： 计算机辅助诊断（CAD）已经广泛地应用于许多疾病的诊断中，利用计算机辅助诊断技术医生可以在不影响诊断效果的前提下提高疾病的诊断效率，从而为病患赢得就医的时间。提出了一种基于组合学习方式的肺栓塞辅助诊断分类器：首先基于AdaBoost训练思想，以CART（分类与回归树）作为弱分类器，构造了一个AdaBoost分类器，然后结合BP神经网络分类器，设计出组合分类器。该分类器不仅提高了诊断的效率，同时也使得诊断效果有了一定的改善。

关键词: 组合学习, 计算机辅助诊断, 肺栓塞, 神经网络, AdaBoost算法