Reduction of interpolation artifacts introduced by moving histogram equalization (MHE) in digital on-line portal images

Digital on-line portal imaging provides almost real time images of patients undergoing radiation treatments, and is used to verify treatment field size and shape, and to achieve greater set-up and treatment accuracy. A major problem with this new modality is the lack of contrast of anatomical structures due to the relatively small differences in attenuation coefficients of bone and soft tissue, and the large amount of scatter at megavoltage photon energies. One solution to this problem is to process the images using adaptive contrast enhancement techniques, one example of which is known as moving histogram equalization (MHE). A disadvantage of MHE is that artifacts often appear in the enhanced image, especially in regions of relative uniformity. A technique has been developed to remove these artifacts while retaining the high contrast of the enhanced image using matched Fourier domain filtering techniques. The artifact reduction factor is found to be about 0.6 for a variety of on-line portal images.