Article

Simulation study on potential accuracy gains from dual energy CT tissue segmentation for low-energy brachytherapy Monte Carlo dose calculations.

Department of Radiation Oncology, GROW School for Oncology and Developmental Biology, Maastricht University Medical Center, Maastricht 6201 BN, The Netherlands.
Physics in Medicine and Biology (Impact Factor: 2.7). 09/2011; 56(19):6257-78. DOI: 10.1088/0031-9155/56/19/007
Source: PubMed

ABSTRACT This work compares Monte Carlo (MC) dose calculations for (125)I and (103)Pd low-dose rate (LDR) brachytherapy sources performed in virtual phantoms containing a series of human soft tissues of interest for brachytherapy. The geometries are segmented (tissue type and density assignment) based on simulated single energy computed tomography (SECT) and dual energy (DECT) images, as well as the all-water TG-43 approach. Accuracy is evaluated by comparison to a reference MC dose calculation performed in the same phantoms, where each voxel's material properties are assigned with exactly known values. The objective is to assess potential dose calculation accuracy gains from DECT. A CT imaging simulation package, ImaSim, is used to generate CT images of calibration and dose calculation phantoms at 80, 120, and 140 kVp. From the high and low energy images electron density ρ(e) and atomic number Z are obtained using a DECT algorithm. Following a correction derived from scans of the calibration phantom, accuracy on Z and ρ(e) of ±1% is obtained for all soft tissues with atomic number Z ∊ [6,8] except lung. GEANT4 MC dose calculations based on DECT segmentation agreed with the reference within ±4% for (103)Pd, the most sensitive source to tissue misassignments. SECT segmentation with three tissue bins as well as the TG-43 approach showed inferior accuracy with errors of up to 20%. Using seven tissue bins in our SECT segmentation brought errors within ±10% for (103)Pd. In general (125)I dose calculations showed higher accuracy than (103)Pd. Simulated image noise was found to decrease DECT accuracy by 3-4%. Our findings suggest that DECT-based segmentation yields improved accuracy when compared to SECT segmentation with seven tissue bins in LDR brachytherapy dose calculation for the specific case of our non-anthropomorphic phantom. The validity of our conclusions for clinical geometry as well as the importance of image noise in the tissue segmentation procedure deserves further experimental investigation.

0 Bookmarks
 · 
97 Views
  • [Show abstract] [Hide abstract]
    ABSTRACT: Purpose: To investigate methods of generating accurate patient-specific computational phantoms for the Monte Carlo calculation of lung brachytherapy patient dose distributions.Methods: Four metallic artifact mitigation methods are applied to six lung brachytherapy patient computed tomography (CT) images: simple threshold replacement (STR) identifies high CT values in the vicinity of the seeds and replaces them with estimated true values; fan beam virtual sinogram replaces artifact-affected values in a virtual sinogram and performs a filtered back-projection to generate a corrected image; 3D median filter replaces voxel values that differ from the median value in a region of interest surrounding the voxel and then applies a second filter to reduce noise; and a combination of fan beam virtual sinogram and STR. Computational phantoms are generated from artifact-corrected and uncorrected images using several tissue assignment schemes: both lung-contour constrained and unconstrained global schemes are considered. Voxel mass densities are assigned based on voxel CT number or using the nominal tissue mass densities. Dose distributions are calculated using the EGSnrc user-code BrachyDose for (125)I, (103)Pd, and (131)Cs seeds and are compared directly as well as through dose volume histograms and dose metrics for target volumes surrounding surgical sutures.Results: Metallic artifact mitigation techniques vary in ability to reduce artifacts while preserving tissue detail. Notably, images corrected with the fan beam virtual sinogram have reduced artifacts but residual artifacts near sources remain requiring additional use of STR; the 3D median filter removes artifacts but simultaneously removes detail in lung and bone. Doses vary considerably between computational phantoms with the largest differences arising from artifact-affected voxels assigned to bone in the vicinity of the seeds. Consequently, when metallic artifact reduction and constrained tissue assignment within lung contours are employed in generated phantoms, this erroneous assignment is reduced, generally resulting in higher doses. Lung-constrained tissue assignment also results in increased doses in regions of interest due to a reduction in the erroneous assignment of adipose to voxels within lung contours. Differences in dose metrics calculated for different computational phantoms are sensitive to radionuclide photon spectra with the largest differences for (103)Pd seeds and smallest but still considerable differences for (131)Cs seeds.Conclusions: Despite producing differences in CT images, dose metrics calculated using the STR, fan beam + STR, and 3D median filter techniques produce similar dose metrics. Results suggest that the accuracy of dose distributions for permanent implant lung brachytherapy is improved by applying lung-constrained tissue assignment schemes to metallic artifact corrected images.
    Medical Physics 01/2014; 41(1):011712. · 2.91 Impact Factor
  • Source
    [Show abstract] [Hide abstract]
    ABSTRACT: Seit kurzem stehen Präzisions-Bestrahlungsgeräte mit einer integrierten, hoch auflösenden Röntgen-CT-Bildgebung für präklinische Studien zur Verfügung. Diese Forschungsplattformen bieten erhebliche Vorteile gegenüber Tier-Bestrahlungsgeräten der älteren Generationen hinsichtlich der Genauigkeit der bildgeführten, gezielten Strahlentherapie. Diese Plattformen werden wahrscheinlich eine entscheidende Rolle bei der Entwicklung von Experimenten spielen, welche die Übertragung von Forschungsergebnissen in klinische Situationen zum Ziel haben. Innerhalb des Fachgebietes Strahlentherapie, aber auch in anderen Bereichen wie zum Beispiel der Neurologie, bieten diese Geräte einzigartige Möglichkeiten, unter anderen Substanzen die Synergie zwischen Bestrahlung und Medikamenten oder anderen Agentien zu erforschen. Um die Vorteile dieser neuen Technologie voll aus-schöpfen zu können, sind genaue Methoden notwendig, um die Bestrahlung planen und die dreidimensionale Dosisverteilung im Organismus berechnen zu können. Spezielle, hierfür entworfene Bestrahlungsplanungssysteme sind hierbei essentiell. In dieser Übersichtsarbeit erörtern wir die spezielle Situation der Präzisionsbestrahlung von Kleintieren, wir beschreiben die Arbeitsweise der Bestrahlungsplanung bei Tieren, und wir untersuchen verschiedene Algorithmen zur Dosisberechnung (Ray Tracing, Superposition-Konvolution, Monte-Carlo-Simulation), die für die Tierbestrahlung mittels Kilovolt-Photonen verwendet werden. Des Weiteren werden Punke, wie zum Beispiel Methoden der Dosismeldung, Photonenstreuung, Gewebesegmentation und Bewegung kurz angerissen.
    Zeitschrift für Medizinische Physik 01/2014; · 1.21 Impact Factor
  • [Show abstract] [Hide abstract]
    ABSTRACT: Dual-energy computed tomography (DECT) images can undergo a two-material decomposition process which results in two images containing material density information. Material density images obtained by that process result in images with increased pixel noise. Noise reduction in those images is desirable in order to improve image quality. A noise reduction algorithm for material density images was developed and tested. A three-level wavelet approach combined with the application of an anisotropic diffusion filter was used. During each level, the resulting noise maps are further processed, until the original resolution is reached and the final noise maps obtained. Our method works in image space and, therefore, can be applied to any type of material density images obtained from any DECT vendor. A quantitative evaluation of the noise-reduced images using the signal-to-noise ratio (SNR), contrast-to-noise ratio (CNR) and 2D noise power spectrum was done to quantify the improvements. The noise reduction algorithm was applied to a set of images resulting in images with higher SNR and CNR than the raw density images obtained by the decomposition process. The average improvement in terms of SNR gain was about 49 % while CNR gain was about 52 %. The difference between the raw and filtered regions of interest mean values was far from reaching statistical significance (minimum [Formula: see text], average [Formula: see text]). We have demonstrated through a series of quantitative analyses that our novel noise reduction algorithm improves the image quality of DECT material density images.
    International Journal of Computer Assisted Radiology and Surgery 05/2014; · 1.36 Impact Factor

Full-text

Download
1 Download
Available from
Nov 25, 2014