Image Quality Assessment, Denoising and Comparative Analysis using Filters for C-arm X-ray Images

C-arm is a fluoroscopy based medical imaging device, which has various operational parameters (dose levels, power applied and frame rate) that has some effect on the image quality. X-ray fluoroscopy exhibits various types of noise (quantum, Gaussian, speckle, photon, Poisson noise, etc.) that must be reduced in real time. Presence of noise degrades the image quality and thus decreases the performance of subsequent application. We have developed a method for statistical quality assessment and denoising technique for flat field detected C-arm X-ray image sequence. The analysis approach is tested on 84 x-ray image sequences acquired from homogeneous phantoms (PMMA) of different widths. Statistical parameters (mean, standard deviation and signal to noise ratio) are considered for generating noise profile. These generated noise profiles are used to study effect on noise due to different widths of phantom, dose levels, applied kV power, and frame rate considered for respective image sequence. Denoising of these image sequence is done using spatial domain filters. Conventional mean, Gaussian and median filters of different kernels are chosen for denoising of image sequence. Comparative results of different filter performance is also studied.