Adaptive Autoregressive Model for Reduction of Poisson Noise in Scintigraphic Images

Edge-preserving adaptive autoregressive model for Poisson noise reduction

Autoregressive models in image processing are linear prediction models that split an image into a predicted (i.e. filtered) image and a prediction error image, which extracts data on the image edges. Edge separation is a crucial feature of an autoregressive model. Data on the edges can be processed in different ways and then added to the filtered image. Another basic feature of our method is spatially varying modelling. In this short article, we propose an improved autoregressive model that preserves image sharpness around the edges of the image and focus on the reduction of Poisson noise, which degrades nuclear medicine images and presents a special challenge in medical imaging.

Comparison of noise estimation methods used in denoising 99mTc-sestamibi parathyroid images using wavelet transform

The objective of this study was to compare the performance of variance, median absolute deviation, and the square of median absolute deviation methods of noise estimation in denoising of ^99mTc-sestamibi parathyroid images using wavelet transform. Sixty-eight ^99mTc-sestamibi parathyroid images including 33 images acquired at zoom 1.0 and 35 acquired at zoom 2.0 were denoised using the wavethresh package in R. The image decomposition and reconstruction method discrete wavelet transform, wavelet filter db4, shrinkage method hard, and thresholding policy universal were used. The noise estimation in the process was made using var, mad and madmad functions, which use variance, mean absolute deviation, and the square of mean absolute deviation, respectively. The quality of denoised images was assessed both qualitatively and quantitatively. A nonparametric two-sample Kolmogorov–Smirnov test was applied to find whether the difference in image quality produced by these three noise estimation methods was significant at 95% confidence. Noise estimation using madmad function produced the best quality denoised image. Further, the quality of the denoised image using madmad function was significantly better than the quality of the denoised image obtained with var or mad function (P = 1). The estimation of noise using madmad functions in wavelet transforms provides the best-denoised image for both zoom 1.0 and zoom 2.0 ^99mTc-sestamibi parathyroid images.

Adaptive Autoregressive Model for Reduction of Noise in SPECT

This paper presents improved autoregressive modelling (AR) to reduce noise in SPECT images. An AR filter was applied to prefilter projection images and postfilter ordered subset expectation maximisation (OSEM) reconstruction images (AR-OSEM-AR method). The performance of this method was compared with filtered back projection (FBP) preceded by Butterworth filtering (BW-FBP method) and the OSEM reconstruction method followed by Butterworth filtering (OSEM-BW method). A mathematical cylinder phantom was used for the study. It consisted of hot and cold objects. The tests were performed using three simulated SPECT datasets. Image quality was assessed by means of the percentage contrast resolution (CR%) and the full width at half maximum (FWHM) of the line spread functions of the cylinders. The BW-FBP method showed the highest CR% values and the AR-OSEM-AR method gave the lowest CR% values for cold stacks. In the analysis of hot stacks, the BW-FBP method had higher CR% values than the OSEM-BW method. The BW-FBP method exhibited the lowest FWHM values for cold stacks and the AR-OSEM-AR method for hot stacks. In conclusion, the AR-OSEM-AR method is a feasible way to remove noise from SPECT images. It has good spatial resolution for hot objects.