STUDY OF POISSON NOISE REDUCTION ON GAMMA CAMERA IMAGE USING SPATIAL DOMAIN FILTER

Abstract

A gamma camera image is produced by a gamma camera that detects the gamma radiation emitted by the radioactive substance or radiopharmaceutical injected into the body. The gamma camera image sometimes has noise that can interfere with the diagnosis. This image is commonly affected by a Poisson-type random noise. This research proposes using a spatial domain filter to study Poisson noise reduction in gamma camera images. The image sample used is the image of a mouse injected with Lu-177-DOTA Trastuzumab with 100 µCi activity detected using a dual-head gamma camera with NaI(Tl) detectors. The grayscale image is treated with Poisson noise, then improved using a spatial domain filter. The spatial domain filters used include Mean, Median, Wiener, and Spatial Lowpass Filters. The mean filter is the best one that can reduce Poisson noise among the four applied filters. The best filter size for noise reduction is 3 with MSE 5.07, PSNR 41.08 dB, and SSIM 0.99.