Medical image analysis relies heavily on picture denoising. Improved perceptual quality of noisy image samples can speed up the diagnostic procedure in many circumstances. Even though medical image denoising is widely used, the current approaches fail to address the wide spectrum of noise in multidisciplinary medical imaging. In this work, new image denoising will be performed by utilizing Bayesian thresholding with LSTM. Here, 5x5 windowing will be applied in the DWT domain to classify the noise effected coefficients. The LSTM will be trained by combining noisy and regular blocks with manually selected label information. The performance will be evaluated using image denoising performance metrics such as 53.25 of PSNR, 0.98 of SSIM, 0.018 of RMSE etc.

Medical Image Denoising, Bayesian Thresholding, Discrete Wavelet Transform, AlexNet, Deep Learning, Peak Signal to Noise Ratio.