Image denoising in volume rendering

In this study, output images produced by medical imaging systems, which have an important role in the diagnosis and treatment of the disease, are used. If the noises of the images used in the medical field are not sufficiently reduced, the use of these images as a medical purpose and diagnosis of the disease becomes very difficult. For this purpose, a new machine learning approach, Extreme Learning Machines (ELM), has been applied for determining noisy pixels in medical images and denoising. The reason behind selecting this method is not only for denoising pixels, but also maintaining critical structural information that can be used for disease diagnosis. The classification techniques (ANN and SVM) with filtering methods (Gauss, NLMF and Kuwahara) are compared with our study. After filtering, some commonly used statistical parameters (MSE, PSNR and SSIM) are calculated to measure the similarity between filtered and original images. The effect of the ELM method on medical images was compared by using multiple computerized tomography (CT) images. The results showed that this method is statistically effective and produces more accurate and higher success rates than other methods. Key Words: Medical imaging, noise detection, extreme learning machines, denoising, filtering