Nonlinear order statistic filters for image filtering and edge detection
Signal Processing
Digital Image Restoration
| Hi-index | 0.00 |
A novel adaptive neural network is proposed for image restoration using a nuclear medicine gamma camera based on the point spread function ofmeasured system. The objective is to restore image degradation due tophoton scattering and collimator photon penetration with the gamma cameraand allow improved quantitative external measurements of radionuclidesin-vivo. The specific clinical model proposed is theimaging of bremsstrahlung radiation using ^32P and^90Y because of the enhanced image degradation effects ofphoton scattering, photon penetration and poor signal/noise ratio inmeasurements of this type with the gamma camera. This algorithm modelavoids the common inverse problem associated with other image restoration filters such as the Wiener filter. The relative performance of the adaptiveNN for image restoration is compared to a previously reported orderstatistic neural network hybrid (OSNNH) filter by these investigators, atraditional Weiner filter and a modified Hopfield neural network usingsimulated degraded images with different noise levels. Quantitative metricssuch as the change of signal to noise ratio (ΔSNR) are used tocompare filter performance. The adaptive NN yields comparable results forimage restoration with a slightly better performance for the images withhigher noise level as often encountered in bremsstrahlung detection withthe gamma camera. Experimental attenuation measurements were also performedin a water tank using two radionuclides, ^32P and ^90Y, typically used for antibody therapy. Similar values foran effective attenuation coefficient was observed for the restored imagesusing the OSNNH filters and adaptive NN which demonstrate that therestoration filters preserves the total counts in the image as required forquantitative in-vivo measurements. The adaptive NN was computationally moreefficient by a factor 4–6 compared to the OSNNH filter. The filterarchitecture, in turn, is also optimum for parallel processing or VLSIimplementation as required for planar and particularly for tomographicmode of detection using the gamma camera. The proposed adaptive NN methodshould also prove to be useful for quantitative imaging of single photonemitters for other nuclear medicine tomographic imaging applications usingpositron emitters and direct X-ray photon detection.