A Novel Speckle Noise Reduction in Biomedical Images using PCA and Wavelet Transform

Medical image processing is used for the diagnosis of diseases by the physicians or radiologists. Noise corrupts medical images and hence qualities of the images are degraded. This degradation includes suppression of edges, structural details, blurring boundaries etc. Therefore image denoising is a very important task and noise should be filtered out, without affecting important features of the image. PCA is a correlation method used to find out the new data smaller than the original variables and retain most of the sample information. The performances of wavelet PCA is further improved by incorporating the intra and inter scale dependencies relation among the wavelet coefficients. For that a multi level representation is performed for the ultrasound image using the stationary wavelet transform and inverse stationary wavelet transform. Feature vectors for each pixel is extracted by using intra and inter scale dependencies data from the stationary wavelet transform. The principal component for each feature vector is obtained by using PCA technique. A soft thresholding function and an improved threshold are applied in the PCA domain to remove noise. The inverse principal component analysis is then applied for reconstruction of the ultrasound image. The proposed denoising filter has better improvement in the peak signal to noise ratio and the edge preservation ability.