Analysis Techniques For Eliminating Noise In Medical Images Using Bivariate Shrinkage Method

Image denoising is a challenging process in digital image processing aiming at the removal of noise and is still a demanding problem for researchers. During acquisition and transmission, images are often corrupted and denoising is an essential step to improve the image quality. In medical imaging like CT-Scan, Magnetic Resonance Image (MRI), EEG, ECG accurate diagnosis is to be done. In Magnetic Resonance Imaging (MRI) images are typically corrupted with noise, which hinder the medical diagnosis based on these images. Noise can be introduced in an image by capturing instruments, data transmission media, image quantization and discrete sources of radiation. Noise degrades the quality of images, suppressing structural details, thus create difficulties to medical diagnosis. Therefore, in medical image de-noising it is necessary to remove the noise while preserving important features. The medical imaging techniques play vital role in providing important information about the organ to the physician in a non invasive manner and help in detecting the disease as early as possible. Practically acquisition time in medical imaging is limited due to patient comfort and system requirement. Therefore, fast imaging is needed. But when the time resolution is improved, the noise may degrades the quality of images, blurring boundaries and suppressing structural details, thus bring difficulties to medical diagnosis. There are various techniques for noise removal from images like Wiener filter, Median filter, Average filter and Wavelet thresholding. Each technique has its assumptions, merits, and demerits. Our proposed method separates unknown signal sources without any prior knowledge. In this paper, we used wavelet based bivariate shrinkage method algorithm as denoising technique and compare its results with existing Wavelet Denoising. Performance results are evaluated in terms of metrics called Peak Signal-to-Noise Ratio (PSNR). Since noise in MR images is non gaussian, results show that proposed technique is a very appropriate analysis technique for eliminating noise in Medical images specially MRI.