为解决基于二维图像处理的计算机辅助诊疗系统(CAD)仅考虑每幅图像自身包含的信息而忽略不同扫描层之间的联系，以及数据处理过程中的海量计算问题，提出一种新的基于快速三维主成分分析(3D PCA)的有效肺CT病灶检测算法。该算法首先引入高维张量奇异值分解(HOSVD)设计3D PCA; 然后以提取出的三维空间特征点为种子点，进行区域增长以获取完整的疑似病灶区域; 最后，根据医学图像具体特征，设计了一种HOSVD的简化分解算法。对来自吉林省肿瘤医院的10个典型病例的五百余幅临床CT图像进行了实验，并将实验结果与当前同类算法做了比较。结果表明，检测精确度提高了约10%~21%; 另外，快速算法与原算法比较，计算复杂度可降低约1/3。由于快速3D PCA可以挖掘更多存在于不同连续扫描层间的有用信息，更精准提取病灶特征，在一定程度上提高了检测率。

To solve the problem that the current Computer Aided Diagnosis(CAD) based 2D schemes only processes the formation for each scanned image itself and ignors the relation between the images, moreover, to reduce the computational complexity of high-order technology in an actual project, a more effective algorithm to detect pulmonary lesions in CT scanned images based on the fast Three Dimension Principle Component Analysis (3DPCA) was presented. Firstly, the Higher-Order Singular Value Decomposition (HOSVD) was introduced to design the 3D PCA. Then, by choosing the feature points as the seed points, the region grow was used to obtain the whole suspected lesion. Finally, a fast decomposition algorithm was presented according to the feature of CT scanned images. The technique was tested against more than 500 CT images form 10 typical cases of Jilin Tumor Hospital. The results confirm the validity of technique as well as enhanced performance. Comparing with other algorithms, the ratio of detection is improved by 10%-21%, and the computation is reduced to 1/3 of the original algorithm. Obtained data show that the fast 3D PCA could excavate more information exiting in the successive CT images. Moreover, improved high order technology can be more effectively applied in actual projects.