Constrained reconstructions for 4D intervention guidance.

Image-guided interventions are an increasingly important part of clinical minimally invasive procedures. However, up to now they cannot be performed under 4D (3D + time) guidance due to the exceedingly high x-ray dose. In this work we investigate the applicability of compressed sensing reconstructions for highly undersampled CT datasets combined with the incorporation of prior images in order to yield low dose 4D intervention guidance. We present a new reconstruction scheme prior image dynamic interventional CT (PrIDICT) that accounts for specific image features in intervention guidance and compare it to PICCS and ASD-POCS. The optimal parameters for the dose per projection and the numbers of projections per reconstruction are determined in phantom simulations and measurements. In vivo experiments in six pigs are performed in a cone-beam CT; measured doses are compared to current gold-standard intervention guidance represented by a clinical fluoroscopy system. Phantom studies show maximum image quality for identical overall doses in the range of 14 to 21 projections per reconstruction. In vivo studies reveal that interventional materials can be followed in 4D visualization and that PrIDICT, compared to PICCS and ASD-POCS, shows superior reconstruction results and fewer artifacts in the periphery with dose in the order of biplane fluoroscopy. These results suggest that 4D intervention guidance can be realized with today's flat detector and gantry systems using the herein presented reconstruction scheme.

The M10 locus of murine gammaherpesvirus 68 contributes to both the lytic and the latent phases of infection.

Murine gammaherpesvirus 68 (MHV-68) is closely related to Epstein-Barr virus and Kaposi's sarcoma-associated herpesvirus (KSHV) and provides a small-animal model to study the pathogenesis of gammaherpesvirus (gammaHV) infections. According to the colinear organization of the gammaHV genomes, the M10 locus is situated at a position equivalent to the K12 locus of KSHV, which codes for proteins of the kaposin family. The M10 locus of MHV-68 has been predicted to code for three overlapping open reading frames (M10a, M10b, and M10c [M10a-c]) with unknown function. In addition, the M10 locus contains a lytic origin of replication (oriLyt). To elucidate the function of the M10 locus during lytic and latent infections, we investigated, both in vitro and in vivo, the following four recombinant viruses which were generated using MHV-68 cloned as a bacterial artificial chromosome: (i) a mutant virus with a deletion which affects both the coding region for M10a-c and the oriLyt; (ii) a revertant virus in which both the M10a-c coding region and the oriLyt were reverted to those of the wild type; (iii) a virus with an ectopic insertion of the oriLyt, which restores the function of the oriLyt but not the M10a-c coding region; and (iv) a mutant virus with a deletion in the oriLyt only. While the mutants were slightly attenuated with regard to lytic replication in cell culture, they showed severe growth defects in vivo. Both lytic replication and latency amplification were strongly reduced. In contrast, both the revertant virus and the virus with the ectopic oriLyt insertion grew very similarly to the parental wild-type virus both in vitro and in vivo. Thus, we provide genetic evidence that mutation of the oriLyt, and not of putative protein coding sequences within the M10a-c region, is responsible for the observed phenotype. We conclude that the oriLyt in the M10 locus plays an important role during infection of mice with MHV-68.

TRLFS: analysing spectra with an expectation-maximization (EM) algorithm.

A new approach for fitting statistical models to time-resolved laser-induced fluorescence spectroscopy (TRLFS) spectra is presented. Such spectra result from counting emitted photons in defined intervals. Any photon can be described by emission time and wavelength as observable attributes and by component and peak affiliation as hidden ones. Understanding the attribute values of the emitted photons as drawn from a probability density distribution, the model estimation problem can be described as a statistical problem with incomplete data. To solve the maximum likelihood task, an expectation-maximization (EM) algorithm is derived and tested. In contrast to the well known least squares method, the advantage of the new approach is its ability to decompose the spectrum into its components and peaks using the revealed hidden attributes of the photons as well as the ability to decompose a background-superimposed spectrum into the exploitable signal of the fluorescent chemical species and the background. This facilitates new possibilities for evaluation of the resulting model parameters. The simultaneous detection of temporal and spectral model parameters provides a mutually consistent description of TRLFS spectra.