Texture characteristic extraction of medical images based on pyramid structure wavelet transform

In order to extract image texture features of medical images, pyramid structure wavelet (PWT) is used in this paper. A group of gray-scale images of chest CT from medical image database are selected as test set. The PWT using three different wavelets bases db6, db2 and Haar are carried out to extract texture feature of each image respectively. A method called symmetric circular extension which takes the boundary as symmetric center are utilized to overcome data increasing and large errors existed nearby reconstructed signal. Empirical results show that the texture feature of chest CT images is mainly concentrated in the low frequency part. However, LH, HL and HH sub-bands contain only less than 7% of the total energy. The PWT using haar wavelet basis extracts more information than it using db2 and db6 wavelet basis. For texture analysis of medical images, the PWT algorithm using haar wavelet base can improve the performance of feature extraction of each sub-band.