Pixel intensity based noise reduction method for single photon emission computed tomography images

A technique for reducing noise in single photon emission computed tomography image is proposed. A variance stabilizing transformation is applied on the noisy image so that Poisson noised image is converted into an approximate additive noise distributed image. De-noising is achieved in the transformed domain using the filter transfer function estimated based on the local image statistics. The extent of smoothing is decided by the pixel intensity and edge information of the image region. This helps in preserving edges. The image is then converted back into the original domain using an appropriate inverse transform. The simulation results show that the proposed method gives better de-noising results in terms of qualitative and quantitative performance measures and outperforms existing techniques in de-noising single photon emission computed tomography images.