The qualities of images obtained in scintigraphic devices are limited because of poor statistics and finite resolving power of the devices and collimators. Thus, processing of the image data such as smoothing, image enhancement, reduction of high frequency noise, boundary detection, etc. may improve information transfer to the human viewers. Improvement of the quality of nuclear medicine images has been attempted by using various filtering techniques. They can be classified into two types: the space-domain filtering and the frequency-domain filtering. This paper analyzed the usefulness and limitations of the technique using the space-domain filtering by comparing the calculation results of entropy in the images processed by various digital filters.

View full abstract

The three-dimensional projection reconstruction is to reconstruct a three-dimensional image, or a three-variable function to be more general, from a set of its plane integrals onto lines. A projection refers to such an integral. This reconstruction problem arises for instance in zeugmatographic, magnetic resonance imaging. A method based on the inverse Radon transform is already known for the three-dimensional reconstruction. However, the quality of the images reconstructed by numerical computation of this transform has not been made clear. We have studied the image property by computer simulation. We reconstructed spherical objects and studied the directionality and an approximation to the point spread function of the method.

View full abstract

The purpose of this work is to materialize real time restoration of the image degraded by signalindependent Gaussian noise. A new method proposed here is based on one-dimensional recursive estimator known as the Kalman filter to reduce processing time. This method is a kind of nonlinear smoothing filter which preserves the sharpness of edge images. Because the noise at the edge is not conspicious, the estimated value at a pixel is regarded as the observed value at the pixel if the difference between the estimated and the observed values is larger than the standard deviation of noise. This new filter preserves sharpness of image better than ordinary smoothing method.

View full abstract