This paper presents the results obtained from an experimental study on the effectiveness of a noise reduction technique applied to medical images. This study compares the effectiveness of various noise-removal filters in order to determine which is the most efficient. Mode, median, Gaussian, adaptive median, non-local mean, wavelet, and deep learning-based denoising convolutional neural network (CNN) filters were used. This paper seeks to demonstrate experimental results on two types of noisy input images used: impulse noise and Gaussian noise, and compare them to the denoising effect of various methods. Among all the methods evaluated, the deep learning-based Convolutional Neural Network (CNN) approach demonstrates superior performance compared to the other techniques. While maintaining the edges and fine structures in the images, this method was able to eliminate both additive and multiplicative noise. The evaluation of these methods performance primarily relied on several metrics, including visual assessment, Mean Square Error (MSE), Peak Signal-to-Noise Ratio (PSNR), Structural Similarity Index (SSIM), Normalized Absolute Mean (NAM), and Cross-Correlation (CC). These metrics provided a comprehensive analysis of the effectiveness and quality of the methods in reducing noise and preserving image details.