A Novel Medical Image De-noising Algorithm for Efficient Diagnosis in Smart Health Environment

Smart healthcare is defined by the technology that leads to better diagnostic tools, better treatment for patients, and devices that improve the quality of life for anyone and everyone. Medical images have significant to facilitate that smart health environment. However, the medical images frequently get noisy in the acquisition process, which engages many different physical mechanisms. Most of the de-noising algorithms conceive the additive white Gaussian noise (AWGN). However, among the popular medical image modalities, several are degraded by some type of non-Gaussian noise, such as Poisson noise. Poisson noise is mainly associated with many imaging modalities like single-photon emission computerized tomography (SPECT), (positron emission tomography) PET, and fluorescent confocal microscopy imaging. Because of the signal-dependent nature of Poisson noise, the various de-noising filters proposed in the literature, including the Non-Local Mean (NL-Mean) filter. In literature, NL-Mean is mostly applied for Gaussian noise extraction and very rarely used for Poisson noise removal. In this work, notable efforts are put to modified NL-Mean filter, and high order NL-Mean Methods are proposed. These novel high order algorithms de-noise images by prominent the signals and noise because it takes the high order odd moment of the medical image. The visual quality of the de-noised medical image (PET) and correlation graph determines that the proposed algorithms outperform the conventional de-noising filter. This study's findings will significantly contribute to the development of a more accurate and robust image analysis model, which is the need of today's modern age of digitization.