Abstract: It studies the application of PCA to the separation of multiple tracers in PET tumour images. First, up-to-95% contribution rate principal components were extracted from two different tracers under noisy circumstances, and then an equation set was established, based on time difference between different tracers and their principal components. and injection -time delay and principal components of the tracers. Then, the coefficient of principal components was obtained via linear least square. Finally, three kinds of tracer curves were attained.  Experimental results show that the relative data are feasible, and the consistency between the result and raw data comes to a satisfactory 80%.

Key words: principal component analysis（PCA）； positron emission tomography（PET)； linear  least square； multiple tracer； wavelet analysis