Medical Image Denoising and Brain Tumor Detection Using CNN and U-Net

The process of medical imaging, which incorporates images produced by X-rays, ultrasound imaging, angiography, etc., is intricate. During the imaging process, image noise is also captured, some of which are very corrosive, causing a disturbance that impairs the quality of the image. The challenge of removing destructive Gaussian additive white noise from medical images while maintaining the fine features is addressed by the suggested approach. The concept of method noise is combined with a deep learning-based framework for a convolutional neural network (CNN), using method noise as a post-processing operation, to create the proposed technique. Brain tumor detection over noisy images is much difficult with better accuracy. Hence in this paper we proposed a method for denoising as well as for brain tumor detection using CNN and U-Net. By explicitly adding Gaussian additive white noise at various noise variance values (= 10, 15, 20, 25, 30), distorted images are produced. The generated denoised images' aesthetic appeal and quantitative metrics, such as peak signal-to-noise ratio (PSNR) and structural similarity index (SSIM), are next assessed. The results are also contrasted with those of other non-traditional methods to evaluate the effectiveness of the suggested methodology further. The findings of the critical examination demonstrate the proposed methodology's admirable effectiveness in degaussing the medical images that had been tainted by Gaussian noise. This method can be applied in a variety of real-world contexts in the realm of medical image processing.