51 Search Results

In this work, we derived explicit relationships between the ideal sinogram and the sinogram degraded by both attenuation and distance-dependent spatial resolution in 2D SPECT that is described by either a Cauchy or a Gaussian function. Attempts to reduce the statistical variance in the reconstructed image lead to the development of infinite classes of closed-form methods for estimation of ideal sinogram. These methods were applied in both computer-simulation and real-data studies.

An infinite class of closed-form methods was developed by the authors last year for image reconstruction in 2D SPECT with uniform attenuation. In the work reported here, they extended their approach to develop a class of closed-form methods that compensate for the effects of both uniform attenuation and distance-dependent spatial resolution in 2D SPECT. These methods, which are characterized by an index n that can be assigned any real number, are exact in the absence of noise but propagate noise differently. They implemented this class of methods for SPECT image reconstruction in both computer-simulation and real-data studies. The results demonstrate that this class of methods corrects effectively for the aforementioned effects. Extensive computer simulation studies indicate that the method obtained with n = 2, which they had proved to be the optimal choice of n in 2D SPECT when only attenuation is present, also provides the smallest global image variance among the methods in the class when compensation for both uniform attenuation and distance-dependent spatial resolution is performed.

A general approach that the authors proposed elsewhere reveals the intrinsic relationship among methods for inversion of the 2-D exponential Radon transform described by Bellini et al., by Tretiak and Metz, by Hawkins et al., and by Inouye et al. Moreover, the approach provides an infinite class of linear methods for inverting the 2-D exponential Radon transform. In the work reported here, they systematically investigated the noise characteristics of the methods in this class, obtaining analytical forms for the autocovariance and the variance of the images reconstructed by use of various methods. The noise properties of a new quasi-optimal method were then compared theoretically to those of other methods of the class. The analysis demonstrates that the quasi-optimal method achieves smaller global variance in the reconstructed images than do the other methods of the class. Extensive numerical simulation studies confirm this prediction.

Exact methods of inverting the two-dimensional (2-D) exponential Radon transform have been proposed by Bellini et al. and by Inouye et al., both of whom worked in the spatial-frequency domain to estimate the 2-D Fourier transform of the unattenuated sinogram; by Hawkins et al., who worked with circularly harmonic Bessel transforms; and by Tretiak and Metz, who followed filtering of appropriately-modified projections by exponentially-weighted backprojection. With perfect sampling, all four of these methods are exact in the absence of projection-data noise, but empirical studies have shown that they propagate noise differently, and no underlying theoretical relationship among the methods has been evident. In this paper, an analysis of the 2-D Fourier transform of the modified sinogram reveals that all previously-proposed linear methods can be interpreted as special cases of a broad class of methods, and that each method in the class can be implemented, in principle, by any one of four distinct techniques. Moreover, the analysis suggests a new member of the class that is predicted to have noise properties better than those of previously proposed members.

The purpose of this work was to evaluate lesion detectability with and without nonuniform attenuation compensation (AC) in myocardial perfusion SPECT imaging in women using an anthropomorphic phantom and receiver operating characteristics (ROC) methodology. Breast attenuation causes artifacts in reconstructed images and may increase the difficulty of diagnosis of myocardial perfusion imaging in women. The null hypothesis tested using the ROC study was that nonuniform AC does not change the lesion detectability in myocardial perfusion SPECT imaging in women. The authors used a filtered backprojection (FBP) reconstruction algorithm and Chang`s single iteration method for AC. In conclusion, with the proposed myocardial defect model nuclear medicine physicians demonstrated no significant difference for the detection of the anterior wall defect; however, a greater accuracy for the detection of the inferior wall defect was observed without nonuniform AC than with it. Medical physicists did not demonstrate any statistically significant difference in defect detection accuracy with or without nonuniform AC in the female phantom.

The aim of this study was to determine which measurement obtained from a radiopharmacokinetic model of a receptor-binding radiotracer provides the highest diagnostic performance for the detectionof diffuse hepatocellular disease. Twenty-seven healthy subjects and 46 patients with diffuse hepatocellular disease were studied with the receptor-binding radiopharmaceutical, {sup 99m}Tc-galactosyl-neoglycoalbumin. A radiopharmacokinetic model was used to produce estimates of receptor concentraiton [R]{sub 0}, the scaled forward-binding rate constant k{sub b}, hepatic plasma volume, V{sub h}, extra-hepatic plasma volume, V{sub e} and hepatic plasma flow, F. Receiver operating characteristic analysis of each model estimate was conducted. Receptor concentration [R]{sub 0} and the metrics [R]{sub 0}/tbw and k{sub b}[R]{sub 0}[R]{sub 0}/tbw provided the best discrimination between healthy an diseased liver. The forward-binding rate constant k{sub b} and the metrices F/V{sub e} and V{sub h}/tbw provided no discrimination. Based on simplicity and higher measurement precision, [R]{sub 0} was selected as the most accurate index of hepatic function.

Various medical imaging modalities offer different but often complementary information. For example, x-ray computed tomography (CT) and magnetic resonance (MR) images provide structural information with high spatial resolution; while positron emission tomography (PET) and single-photon emission computed tomography (SPECT) give functional information with less desirable image quality. Integration of images from multiple modalities opens new avenues for developing innovative image reconstruction algorithms that can provide improved image quality. This paper reports on a Bayesian method for eCT image reconstruction developed to incorporate a priori information derived from the spatially-correlated CT and MR images. These anatomic maps, showing boundaries between regions that exhibit distinctly different characteristics, can be incorporated in the Bayesian method, this improving the spatial resolution and noise properties. The correlated structural information can be used also as templates for deriving correction factors for the effect of photon attenuation, this improving the quantitative accuracy and noise properties. Results from computer simulation studies show significant improvements in image quality.

If the performance of a diagnostic imaging system is to be evaluated objectively and meaningfully, one must compare radiologists' image-based diagnoses with actual states of disease and health in a way that distinguishes between the inherent diagnostic capacity of the radiologists' interpretations of the images, and any tendencies to under-read or over-read. ROC methodology provides the only known basis for distinguishing between these two aspects of diagnostic performance. After identifying the fundamental issues that motivate ROC analysis, this article develops ROC concepts in an intuitive way. The requirements of a valid ROC study and practical techniques for ROC data collection and data analysis are sketched briefly. A survey of the radiologic literature indicates the broad variety of evaluation studies in which ROC analysis has been employed. 162 references.

Recent developments in statistical theory and associated computational techniques have opened new avenues for image reconstruction in positron emission tomography. Algorithms based on maximum likelihood methods have received considerable attention for their potential of providing improved image quality. However, noise properties of those images produced by these methods tend to deteriorate after a certain stage in the iterative process. The authors have developed a Bayesian method that incorporates a priori information to improve the resulting image quality. This approach utilized a Gibbs prior to describe correlation of neighboring regions and takes into account the effect of limited spatial resolution. The Gibbs prior includes features depicting the similarity of intensities in neighboring pixels within homogeneous regions and line sites outlined as boundaries between regions. The effect of limited spatial resolution is incorporated into the probability density functions relating image cells to detector bins. Other physical factors, in principle, can be included as well.

Tomographic image reconstruction requires redistribution of data collected at different projection angles back into the fixed matrix frame. This process normally produces additional noise on the reconstructed images. Three angular reprojection methods have been investigated in the present study. The direct method reprojects the entire rotated pixel (RP) into the fixed pixel (FP) which contains the center of the RP. It is fast but suffers from high noise level. The weighting method redistributes the RP into a number of FPs with weights proportional to the fractional overlaps of these FPs with a shifted FP whose center is that of the RP. This approach reduces the magnitude of the noise but requires more processing time. The matrix method uses matrix coefficients retrieved from a precalculated data bank to transform RPs into FPs. This method introduces least error but demands substantial memory space. Computer simulation studies have been performed to evaluate these three methods. In general, reprojection noise decreases with the use of more projection angles but appears to be independent from both matrix size and object size. Two models have been developed to describe the standard deviation of the noise in reprojected images. For the direct method, this noise can be regarded as a background noise level proportional to count density, which is added to the Poisson image noise. For the weighting and matrix methods the overall effect is as if the Poisson image noise were smoothed by a pixel size 1.5 times as large in each direction as the pixels actually employed, with a background noise level proportional to count density then superimposed. Noise textures also have been studied by Wiener spectral analysis.