F-Test and One-Way ANOVA for Medical Images Diagnosis

A new technique for impulse noise discrimination of medical MRI images is presented. Statistical F-distribution supported by Monte Carlo simulation for prediction of impulse noise location on distorted images is used. The selective noisy pixels are detected as noisy using the One-Way ANOVA hypothesis method. The statistical prediction method is applied to the test images to predict the noise location. The distorted portions in the images predicted by this structure are improved by using a proper nonlinear filter. The nonlinear median filter is showing a suitable way the replacement of degraded pixels. The filter is not applied to the not-detected portions of the image to keep it unchanged and avoid the distortion of fine image details and important edges. Experimental results improved this structure compared with classical noise reduction structures.