arXiv subjects

Krzysztof Kacperski

Publications and source records attributed to Krzysztof Kacperski.

5 recordsLinked to original sources

arXiv2018

High Sensitivity Collimators for Optimising Lesion Detection in SPECT Images

Parallel hole collimators which are currently routinely used in SPECT imaging were designed a few decades ago, when filtered backprojection (FBP) was used for tomographic reconstruction. Statistical reconstruction methods with precise modeling of the projection measurement offer a different standard, and the question of optimizing resolution - sensitivity tradeoff by choosing appropriate collimator aperture needs to be revised. In this paper we search for a parallel hole collimator which offers best performance in detection of hot lesions on a uniform background. To evaluate the image quality we use standard quantitative measures like the contrast recovery coefficient, coefficient of variation, and contrast to noise ratio. We also performed signal detection by human observers followed by Receiver Operating Characteristics (ROC) analysis for a few lesions close to the limit of visibility. Our results consistently indicate that optimal performance is achieved with collimators of sensitivity 6 - 9 times that of the conventional high resolution (HR) collimator. When the optimized collimator is applied, the image quality of the conventional image obtained with the HR can be achieved at scanning time or activity dose reduced by factor of 3.

physics.med-ph↗

arXiv2018

Reduced effective scanning time in SPECT due to OSEM accelerated reconstruction

We study the increase of image noise which can be observed in images reconstructed with the OSEM algorithm. We argue that the excessive noise is equivalent to shortening scanning time by a factor of 2 or more at high number of subsets (acceleration factors). Therefore, by keeping the number of OSEM subsets low one can significantly reduce image noise, which in turn is equivalent to increasing scanning time or radiation dose. While the OSEM remains a very useful reconstruction method enabling substantial reduction of reconstruction times, it should be used with caution. In particular, high numbers of subsets should be avoided if possible. Whenever the available computing power allows achieving acceptable reconstruction times at lower number of subsets, one can consider reducing scanning time or injected dose, while maintaining the image quality of standard scans reconstructed with higher OSEM acceleration. The recommended maximal "safe" number of subsets, which does not lead to significant increase of image noise, is about 5.

physics.med-ph↗

arXiv2005

Performance of three-photon PET imaging: Monte Carlo simulations

We have recently introduced the idea of making use of three-photon positron annihilations in positron emission tomography. In this paper the basic characteristics of the three-gamma imaging in PET are studied by means of Monte Carlo simulations and analytical computations. Two typical configurations of human and small animal scanners are considered. Three-photon imaging requires high energy resolution detectors. Parameters currently attainable by CdZnTe semiconductor detectors, the technology of choice for the future development of radiation imaging, are assumed. Spatial resolution is calculated as a function of detector energy resolution and size, position in the field of view, scanner size, and the energies of the three gamma annihilation photons. Possible ways to improve the spatial resolution obtained for nominal parameters: 1.5 cm and 3.2 mm FWHM for human and small animal scanners, respectively, are indicated. Counting rates of true and random three-photon events for typical human and small animal scanning configurations are assessed. A simple formula for minimum size of lesions detectable in the three-gamma based images is derived. Depending on the contrast and total number of registered counts, lesions of a few mm size for human and sub mm for small animal scanners can be detected.

physics.med-ph↗

arXiv2000

Phase transitions in social impact models of opinion formation

We study phase transitions in models of opinion formation which are based on the social impact theory. Two different models are discussed: (i) a cellular--automata based model of a finite group with a strong leader where persons can change their opinions but not their spatial positions, and (ii) a model with persons treated as active Brownian particles interacting via a communication field. In the first model, two stable phases are possible: a cluster around the leader, and a state of social unification. The transition into the second state occurs for a large leader strength and/or for a high level of social noise. In the second model, we find three stable phases, which correspond either to a ``paramagnetic'' phase (for high noise and strong diffusion), a ``ferromagnetic'' phase (for small nose and weak diffusion), or a phase with spatially separated ``domains'' (for intermediate conditions).

cond-mat.stat-mech↗

arXiv1998

Anomalous oscillations of average transient lifetimes near crises

It is common that the average length of chaotic transients appearing as a consequence of crises in dynamical systems obeys a power low of scaling with the distance from the crisis point. It is, however, only a rough trend; in some cases considerable oscillations can be superimposed on it. In this letter we report anomalous oscillations due to the intertwined structure of basins of attraction. We also present a simple geometrical model that gives an estimate of the period and amplitude of these oscillations. The results obtained within the model coincide with those yielded by computer simulations of a kicked spin model and the Henon map.

chao-dyn↗