In this paper we proposed a hybrid method for Poisson noise amputation in MRI (Magnetic Resonance Imaging) datasets. The awareness should be paid while enhancing the Poisson noise, because it is vastly signal dependent. The Independent Component Analysis and Purelet (Poisson Unbiased Risk Estimation) performances are assessed directly. In purelet we are using Linear Expansion of thresholding. We approximate the mean square error to bring a dutiful transform domain thresholding. The hybrid method is in cooperation with ICA and Purelet Method. Foremost the ICA bring into play intended for dimensionality reduction for Multivariate Data, Subsequently Wavelet threshold Method exploit for Denoising. Performance Comparison is exploited in requisite of PSNR (Peak Signal to Noise Ratio) and speed of Denoising