Respective roles of scatter, attenuation, depth-dependent collimator response and finite spatial resolution in cardiac single-photon emission tomography quantitation: a Monte Carlo study

Abstract

The purpose of this study was to investigate the relative influence of scatter, attenuation, depth-dependent collimator response and finite spatial resolution upon the image characteristics in cardiac single-photon emission tomography (SPET). An acquisition of an anthropomorphic cardiac phantom was performed together with corresponding SPET Monte Carlo simulations. The cardiac phantom and the Monte Carlo simulations were designed so that the effect of scatter, attenuation, depth-dependent collimator response and finite spatial resolution could be studied individually and in combination. The impact of each physical effect and of combinations of effects was studied in terms of absolute and relative quantitative accuracy, spatial resolution and signal-to-noise ratio (SNR) in the resulting images. No corrections for these effects were assessed. Results obtained from Monte Carlo simulations and real acquisitions were in excellent agreement. Attenuation introduced about 90% activity underestimation in a 10-mm-thick left ventricle wall while finite spatial resolution alone introduced about 30% activity underestimation. Scatter had a negligible impact on quantitative accuracy in the recontructed slices when attenuation was present. Neither bull's eye map homogeneity nor contrast between a hot and a cold region were affected by depth-dependent collimator response or finite spatial resolution. Bull's eye map homogeneity was severely affected by attenuation but not by scatter. Attenuation and scatter reduced contrast by about 20% each. Both attenuation and scatter increased the full-width at half-maximum (FWHM) characterizing the spatial resolution of the imaging system by approximately 1 mm each but the main effect responsible for the observed 11-mm FWHM spatial resolution was the depth-dependent collimator response. SNR was reduced by a factor of approximately 2.5 because of attenuation, while scattered counts increased SNR by approximately 10%. In conclusion, the quantification of the relative influence of the different physical effects showed that attenuation is definitely the major phenomenon affecting cardiac SPET imaging accuracy, but that finite spatial resolution, scatter and depth-dependent collimator response also contribute significantly to the errors in absolute and relative quantitation and to the poor spatial resolution.

Imaging studies of biodistribution and kinetics in drug development

“…Then, compensation must be made for the efficiency of the detection device and for spatial variations in detection efficiency caused by losses of radiation due to absorption and Compton scattering in intervening tissues [el-Fakhri et al, 1999;Buvat et al, 1998Buvat et al, , 1995Miller et al, 1996;. Quantification by PET has been extensively validated [Rhodes and Hughes, 1995;Korf, 1997;Leskinen, 1994;Czernin and Schelbert, 1994;Kuwert et al, 1992].…”

“…They are also important in qualitative imaging because of the large effect they have on perceived tracer distribution within the body (note that Figs. 4 and 5 are images of a single drug formulation in a single person), but they are rarely done using sufficiently rigorous methods for the type of data being analyzed [Buvat et al, 1995[Buvat et al, , 1998el-Fakhri et al, 1999;Perring et al, 1994;Lee et al, 1999aLee et al, -c, 2000b. The use of the simplest effective corrections in a planar study introduces a degree of complexity that rivals that of a full PET study.…”

Iterative reconstruction of SPECT data with adaptive regularization

“…Kang and Katsaggelos have shown that it is possible to simultaneously estimate an image and its degree of regularization by making an explicit dependence between them [12]. The simultaneous estimation is obtained by finding the minimum of (6) under the constraint of a linear relationship between the regularization parameter and the regularizing functional (7) By so doing, is made proportional to and , which both increase with the noise in the data. The value of can be set such that the function is monotonically increasing, mapping into and such that the functional is convex with a unique minimum.…”

“…Attenuation correction can thus be performed, which is essential to the quantification of SPECT studies [4]- [6]. Iterative algorithms are compulsory for reconstructing images from SPECT data with the aim of quantification using attenuation correction [7].…”

Diffusion regularization for iterative reconstruction in emission tomography

“…Attenuation is responsible for large biases with underestimation of the distribution values as large as 80% in cardiac SPECT, independently of the size of the structure [2]. Attenuation correction is thus required for quantification purpose rather than resolution recovery.…”