Choice of initial conditions in the ML reconstruction of fan-beam transmission with truncated projection data

Estimation of the Initial Image's Contributions to the Iterative Landweber Reconstruction

This paper develops a closed-form frequency-domain transfer function to predict the initial image's contributions in an iterative Landweber reconstruction for a given number of iterations. The transfer function shows that the initial image's high frequency components play an important role when the iterative algorithm is stopped before convergence. The contributions from the initial image diminish as the iteration number increases if the imaging matrix has full rank.

New reconstruction algorithm, distance dependent exact, with reduced statistical α error, dedicated to emission tomography using a collimator with large and long holes

SPECT (single photon emission computerized tomography) is physically one of the worst medical imaging modalities. Despite a considerable and still increasing medical impact, its spatial resolution (beyond 1 cm) and its sensitivity (less than 10(-4)) are both awful. This situation is mainly due to the use of a thin parallel hole collimator. In addition the application of the unfitted radon-transform worsens the figure.

Optimization of the acquisition geometry in digital tomosynthesis of the breast

Digital tomosynthesis of the breast continues to be intensively studied as an adjunct or replacement of conventional mammography. Although many of the acquisition parameters found in tomosynthesis imaging are also found in conventional mammography and therefore most of the traditional values from mammography have been used in the former, two acquisition geometry parameters, the angular range covered during acquisition and the number of projections per projection set, are applicable only to tomosynthesis. Therefore, in the preclinical and clinical studies reported on tomosynthesis of the breast, a wide variety of values have been used for these two parameters. In this study, 63 different combinations of angular range and number of projections were evaluated using computer simulation methods to characterize how these two parameters affect reconstruction quality and to find which of these combinations maximize it. For this, a computer simulation of a digital tomosynthesis system that included empirically determined system response characteristics was developed and used to generate 9450 different breast tissue volume reconstructions. These reconstructions were analyzed objectively using metrics for in-plane lesion visibility and vertical resolution in the form of the contrast-to-noise ratio and artifact spread function (ASF). It was found that although maximizing the angular range covered does always increase the vertical resolution in tomosynthesis, increasing the number of projections in the acquisition set beyond a relatively low threshold does not further improve the vertical resolution. This threshold value for the minimal number of projections needed to minimize the ASF was found to vary proportionally with angular range. For example, for a 60 degrees angular range, the threshold number of projections was found to be 13. Given the clear inverse relationship between the number of projections and in-plane reconstruction quality under a limited total glandular dose condition, the optimum acquisition geometry in tomosynthesis imaging of the breast is that which maximizes the angular range while maintaining the number of projections close to the threshold values found. Finally, of the 63 acquisition geometries studied, the one that resulted in the highest quality reconstruction, considering both in-plane quality and vertical resolution, consisted of the acquisition of 13 projections over a 60 degrees angular range.

Volume emission rate tomography from a satellite platform

The possibility of retrieving horizontal atmospheric structure from a series of limb images taken aboard a satellite is discussed and a maximum likelihood expectation maximization algorithm is developed. Examples of the retrieval of horizontal structure with this algorithm, for different S/N (signal-to-noise) ratios and different structures, are presented. It is shown that with this algorithm and even in the presence of substantial observational noise, a S/N equal to 10 for a single observation, it is possible to retrieve both horizontal and vertical atmospheric structure.