Robert Brauweiler

Perfusion measurements by micro-CT using prior image constrained compressed sensing (PICCS): initial phantom results

Micro-CT scanning has become an accepted standard for anatomical imaging in small animal disease and genome mutation models. Concurrently, perfusion imaging via tracking contrast dynamics after injection of an iodinated contrast agent is a well-established tool for clinical CT scanners. However, perfusion imaging is not yet commercially available on the micro-CT platform due to limitations in both radiation dose and temporal resolution. Recent hardware developments in micro-CT scanners enable continuous imaging of a given volume through the use of a slip-ring gantry. Now that dynamic CT imaging is feasible, data may be acquired to measure tissue perfusion using a micro-CT scanner (CT Imaging, Erlangen, Germany). However, rapid imaging using micro-CT scanners leads to high image noise in individual time frames. Using the standard filtered backprojection (FBP) image reconstruction, images are prohibitively noisy for calculation of voxel-by-voxel perfusion maps. In this study, we apply prior image constrained compressed sensing (PICCS) to reconstruct images with significantly lower noise variance. In perfusion phantom experiments performed on a micro-CT scanner, the PICCS reconstruction enabled a reduction to 1/16 of the noise variance of standard FBP reconstruction, without compromising the spatial or temporal resolution. This enables a significant increase in dose efficiency, and thus, significantly less exposure time is needed to acquire images amenable to perfusion processing. This reduction in required irradiation time enables voxel-by-voxel perfusion maps to be generated on micro-CT scanners. Sample perfusion maps using a deconvolution-based perfusion analysis are included to demonstrate the improvement in image quality using the PICCS algorithm.

Potential of high-Z contrast agents in clinical contrast-enhanced computed tomography

PURPOSE Currently, only iodine- and barium-based contrast media (CM) are used in clinical contrast-enhanced computed tomography (CE-CT). High-Z metals would produce a higher contrast at equal mass density for the x-ray spectra used in clinical CT. Using such materials might allow for significant dose reductions in CE-CT. The purpose of this study was to quantify the potential for dose reduction when using CM based on heavy metals. METHODS The contrast-to-noise ratio weighted by dose (CNRD) was determined as a function of scan protocol by means of measurements and simulations on a clinical CT scanner. For simulations, water cylinders with diameters 160, 320, 480, and 640 mm were used to cover a broad range of patient sizes. Measurements were conducted with 160 and 320 mm water-equivalent plastic cylinders. A central bore of 13 mm diameter was present in all phantoms. The tube voltage was varied from 80 to 140 kV for measurements and from 60 to 180 kV for simulations. Additional tin filtration of thicknesses 0.4, 0.8, and 1.2 mm was applied in the simulation to evaluate a range of spectral hardness. The bore was filled with a mixture of water and 10 mg/ml of pure iodine, holmium, gadolinium, ytterbium, osmium, tungsten, gold, and bismuth for the simulations and with aqueous solutions of ytterbium, tungsten, gold, and bismuth salts as well as Iopromid containing 10 mg/ml of the pure materials for the measurements. CNRDs were compared to iodine at phantom size-dependent reference voltages for all high-Z materials and the resulting dose reduction was calculated for equal contrast-to-noise ratio. RESULTS Dose reduction potentials strongly depended on phantom size, spectral hardness, and tube voltage. Depending on the added filtration, a dose reduction of 19%-60% could be reached at 80 kV with gadolinium for the 160 mm phantom, 52%-69% at 100 kV with holmium for the 320 mm phantom, 62%-78% with 120 kV for hafnium and the 480 mm phantom and 74%-86% with 140 kV for gold and the 640 mm phantom. While gadolinium might be considered at 160 mm diameter, hafnium showed the best overall performance for phantom sizes of 320 mm and above. The measurements conducted on the clinical CT scanner showed very good agreement with simulations with deviations in the order of 5 to 10%. CONCLUSIONS The results of this study encourage the development and use of CM based on high-Z materials, especially for adipose patients, where high tube voltages are necessary to reach sufficiently short scan times. Hafnium proved to be the best compromise for average-size and for adipose patients. Even higher-Z materials such as gold and bismuth showed a good overall performance in conjunction with high tube voltage, large patients or strong added filtration and may be recommended for scans under these conditions.

Dosimetry concepts for scanner quality assurance and tissue dose assessment in micro‐CT

PURPOSE At present, no established methods exist for dosimetry in micro computed tomography (micro-CT). The purpose of this study was therefore to investigate practical concepts for both dosimetric scanner quality assurance and tissue dose assessment for micro-CT. METHODS The computed tomography dose index (CTDI) was adapted to micro-CT and measurements of the CTDI both free in air and in the center of cylindrical polymethyl methacrylate (PMMA) phantoms of 20 and 32 mm diameter were performed in a 6 month interval with a 100 mm pencil ionization chamber calibrated for low tube voltages. For tissue dose assessment, z-profile measurements using thermoluminescence dosimeters (TLDs) were performed and both profile and CTDI measurements were compared to Monte Carlo (MC) dose calculations to validate an existing MC tool for use in micro-CT. The consistency of MC calculations and TLD measurements was further investigated in two mice cadavers. RESULTS CTDI was found to be a reproducible quantity for constancy tests on the micro-CT system under study, showing a linear dependence on tube voltage and being by definition proportional to mAs setting and z-collimation. The CTDI measured free in air showed larger systematic deviations after the 6 month interval compared to the CTDI measured in PMMA phantoms. MC calculations were found to match CTDI measurements within 3% when using x-ray spectra measured at our micro-CT installation and better than 10% when using x-ray spectra calculated from semi-empirical models. Visual inspection revealed good agreement for all z-profiles. The consistency of MC calculations and TLD measurements in mice was found to be better than 10% with a mean deviation of 4.5%. CONCLUSIONS Our results show the CTDI implemented for micro-CT to be a promising candidate for dosimetric quality assurance measurements as it linearly reflects changes in tube voltage, mAs setting, and collimation used during the scan, encouraging further studies on a variety of systems. For tissue dose assessment, MC calculations offer an accurate and fast alternative to TLD measurements allowing for dose calculations specific to any geometry and scan protocol.

Dynamic contrast-enhanced micro-computed tomography correlates with 3-dimensional fluorescence ultramicroscopy in antiangiogenic therapy of breast cancer xenografts.

ObjectivesDynamic contrast-enhanced (DCE) micro–computed tomography (micro-CT) has emerged as a valuable imaging tool to noninvasively obtain quantitative physiological biomarkers of drug effect in preclinical studies of antiangiogenic compounds. In this study, we explored the ability of DCE micro-CT to assess the antiangiogenic treatment response in breast cancer xenografts and correlated the results to the structural vessel response obtained from 3-dimensional (3D) fluorescence ultramicroscopy (UM). Material and MethodsTwo groups of tumor-bearing mice (KPL-4) underwent DCE micro-CT imaging using a fast preclinical dual-source micro-CT system (TomoScope Synergy Twin, CT Imaging GmbH, Erlangen, Germany). Mice were treated with either a monoclonal antibody against the vascular endothelial growth factor or an unspecific control antibody. Changes in vascular physiology were assessed measuring the mean value of the relative blood volume (rBV) and the permeability-surface area product (PS) in different tumor regions of interest (tumor center, tumor periphery, and total tumor tissue). Parametric maps of rBV were calculated of the tumor volume to assess the intratumoral vascular heterogeneity. Isotropic 3D UM vessel scans were performed from excised tumor tissue, and automated 3D segmentation algorithms were used to determine the microvessel density (MVD), relative vessel volume, and vessel diameters. In addition, the accumulation of coinjected fluorescence-labeled trastuzumab was quantified in the UM tissue scans to obtain an indirect measure of vessel permeability. Results of the DCE micro-CT were compared with corresponding results obtained by ex vivo UM. For validation, DCE micro-CT and UM parameters were compared with conventional histology and tumor volume. ResultsExamination of the parametric rBV maps revealed significantly different patterns of intratumoral blood supply between treated and control tumors. Whereas control tumors showed a characteristic vascular rim pattern with considerably elevated rBV values in the tumor periphery, treated tumors showed a widely homogeneous blood supply. Compared with UM, the physiological rBV maps showed excellent agreement with the spatial morphology of the intratumoral vascular architecture. Regional assessment of mean physiological values exhibited a significant decrease in rBV (P < 0.01) and PS (P < 0.05) in the tumor periphery after anti–vascular endothelial growth factor treatment. Structural validation with UM showed a significant reduction in reduction of relative vessel volume (rVV) (P < 0.01) and MVD (P < 0.01) in the corresponding tumor region. The reduction in rBV correlated well with the rVV (R = 0.73 for single values and R = 0.95 for mean values). Spatial maps of antibody penetration showed a significantly reduced antibody accumulation (P < 0.01) in the tumor tissue after treatment and agreed well with the physiological change of PS. Examination of vessel diameters revealed a size-dependent antiangiogenic treatment effect, which showed a significant reduction in MVD (P < 0.001) for vessels with diameters smaller than 25 &mgr;m. No treatment effect was observed by tumor volume. ConclusionsNoninvasive DCE micro-CT provides valuable physiological information of antiangiogenic drug effect in the intact animal and correlates with ex vivo structural analysis of 3D UM. The combined use of DCE micro-CT with UM constitutes a complementary imaging toolset that can help to enhance our understanding of antiangiogenic drug mechanisms of action in preclinical drug research.

Optical tracking of contrast medium bolus to optimize bolus shape and timing in dynamic computed tomography

One of the biggest challenges in dynamic contrast-enhanced CT is the optimal synchronization of scan start and duration with contrast medium administration in order to optimize image contrast and to reduce the amount of contrast medium. We present a new optically based approach, which was developed to investigate and optimize bolus timing and shape. The time-concentration curve of an intravenously injected test bolus of a dye is measured in peripheral vessels with an optical sensor prior to the diagnostic CT scan. The curves can be used to assess bolus shapes as a function of injection protocols and to determine contrast medium arrival times. Preliminary results for phantom and animal experiments showed the expected linear behavior between dye concentration and absorption. The kinetics of the dye was compared to iodinated contrast medium and was found to be in good agreement. The contrast enhancement curves were reliably detected in three mice with individual bolus shapes and delay times of 2.1, 3.5 and 6.1 s, respectively. The optical sensor appears to be a promising approach to optimize injection protocols and contrast enhancement timing and is applicable to all modalities without implying any additional radiation dose. Clinical tests are still necessary.

Spectral optimization for micro-CT

PURPOSE To optimize micro-CT protocols with respect to x-ray spectra and thereby reduce radiation dose at unimpaired image quality. METHODS Simulations were performed to assess image contrast, noise, and radiation dose for different imaging tasks. The figure of merit used to determine the optimal spectrum was the dose-weighted contrast-to-noise ratio (CNRD). Both optimal photon energy and tube voltage were considered. Three different types of filtration were investigated for polychromatic x-ray spectra: 0.5 mm Al, 3.0 mm Al, and 0.2 mm Cu. Phantoms consisted of water cylinders of 20, 32, and 50 mm in diameter with a central insert of 9 mm which was filled with different contrast materials: an iodine-based contrast medium (CM) to mimic contrast-enhanced (CE) imaging, hydroxyapatite to mimic bone structures, and water with reduced density to mimic soft tissue contrast. Validation measurements were conducted on a commercially available micro-CT scanner using phantoms consisting of water-equivalent plastics. Measurements on a mouse cadaver were performed to assess potential artifacts like beam hardening and to further validate simulation results. RESULTS The optimal photon energy for CE imaging was found at 34 keV. For bone imaging, optimal energies were 17, 20, and 23 keV for the 20, 32, and 50 mm phantom, respectively. For density differences, optimal energies varied between 18 and 50 keV for the 20 and 50 mm phantom, respectively. For the 32 mm phantom and density differences, CNRD was found to be constant within 2.5% for the energy range of 21-60 keV. For polychromatic spectra and CMs, optimal settings were 50 kV with 0.2 mm Cu filtration, allowing for a dose reduction of 58% compared to the optimal setting for 0.5 mm Al filtration. For bone imaging, optimal tube voltages were below 35 kV. For soft tissue imaging, optimal tube settings strongly depended on phantom size. For 20 mm, low voltages were preferred. For 32 mm, CNRD was found to be almost independent of tube voltage. For 50 mm, voltages larger than 50 kV were preferred. For all three phantom sizes stronger filtration led to notable dose reduction for soft tissue imaging. Validation measurements were found to match simulations well, with deviations being less than 10%. Mouse measurements confirmed simulation results. CONCLUSIONS Optimal photon energies and tube settings strongly depend on both phantom size and imaging task at hand. For in vivo CE imaging and density differences, strong filtration and voltages of 50-65 kV showed good overall results. For soft tissue imaging of animals the size of a rat or larger, voltages higher than 65 kV allow to greatly reduce scan times while maintaining dose efficiency. For imaging of bone structures, usage of only minimum filtration and low tube voltages of 40 kV and below allow exploiting the high contrast of bone at very low energies. Therefore, a combination of two filtrations could prove beneficial for micro-CT: a soft filtration allowing for bone imaging at low voltages, and a variable stronger filtration (e.g., 0.2 mm Cu) for soft tissue and contrast-enhanced imaging.

Development and Evaluation of a Phantom for Dynamic Contrast-enhanced Imaging

ObjectivesDynamic contrast-enhanced imaging allows assessing functional information in addition to morphology using various modalities. Several applications have been established in clinical practice; however, there is no standard with respect to injection protocols or postprocessing algorithms. The purpose of this study was to develop a phantom for generating reproducible contrast-enhancement curves and providing a standard for comparison of different protocols and modalities in dynamic imaging. Materials and MethodsOur experimental setup consists of a peristaltic pump to generate a water flow through the phantom and a contrast injection pump. The phantom holds a sequence of layers allowing for assessment of perfusion, signal-to-noise ratio, and spatiotemporal resolution; the latter is the spatial resolution of structures with temporally changing contrast. Reproducibility was evaluated by the functional parameters time to peak, mean transit time, and peak enhancement by 24 scans over 4 weeks on a clinical computed tomography scanner. In addition, the area under the curve was evaluated for different injection durations at constant injection volume. Spatiotemporal resolution was assessed by spatial profiles on perfused bore patterns and compared for standard reconstructions, smooth reconstructions, and highly constrained backprojection for local reconstruction (HYPR LR). ResultsThe phantom showed good reproducibility in repeated measurements, with maximal deviations of 4% for time to peak, 9% for mean transit time, and 8% for peak enhancement. Area under the curve was constant within 3.5% for different injection protocols. For the static case, HYPR LR maintained spatial resolution. For dynamic objects, however, HYPR LR reduced spatial resolution dependent on temporal dynamics by up to 19% for highest dynamics, which was still superior to smooth reconstructions (27%). ConclusionsThe proposed phantom showed good reproducibility and therefore allows for comparing injection protocols or modalities in dynamic imaging. Assessment of spatiotemporal resolution under measurement conditions provides means for assessing postprocessing methods and reconstruction techniques in dynamic imaging.

Time‐delayed summation as a means of improving resolution on fast rotating computed tomography systems

PURPOSE Modern computed tomography (CT) systems are supporting increasingly fast rotation speeds, which are a prerequisite for fast dynamic acquisition, e.g. in perfusion imaging, and for new modalities such as dedicated breast CT, where breathhold scanning is indicated. However, not all detector technologies are supporting the high frame rates that are necessary to retain high resolution for objects far away from the isocenter. Even on systems that would support a sufficiently high frame rate, the necessary bandwidth of the data transfer from the rotating gantry stills remains challenging. The authors evaluated a pixel shifting technique termed time-delayed summation (TDS) as a method of increasing resolution on fast rotating CT systems without the need to increase the frame rate. METHODS In TDS mode, detector pixel values are shifted along rows during image acquisition to compensate for detector motion. In order to fully exploit TDS, focal spot position control (FSC) was used in combination with TDS. FSC applies a counter movement to the x-ray focal spot during image acquisition such that it is kept fixed in space. As a proof of concept, measurements were performed on a prototype photon counting detector capable of TDS. The detector was mounted on a movable table and a gold wire phantom was imaged with different TDS settings and detector velocities. Additionally, simulations of a broad range of TDS and FSC settings on two different modalities, a clinical CT scanner and a breast CT scanner, and two different detector geometries, flat and cylindrical, were performed to assess the gain in resolution and contrast in cylindrical water phantoms containing a small wire at distances from the phantom center varied from 5% to 90% of the phantom radius. As figures of merit, the modulation transfer function (MTF) at 10% and the maximum contrast were used and compared against the respective values when using step-and-shoot acquisition, which means stopping the rotation when a projection image is acquired. RESULTS Measurements showed that detector movement and the resulting blurring of the wire projections were compensated to the expected degree when using the appropriate number of TDS shifts per frame (TDS factor). Using simulations it was found that when using the optimal TDS factor, over 90% of the resolution achieved in step-and-shot mode was reached for all investigated wire positions. TDS showed better performance on a cylindrical detector that on the same system with a flat detector. TDS factors that were deviating from the optimum by more than 1 shift led to a performance below that of standard continuous acquisition. CONCLUSIONS The findings of this study encourage the combined usage of TDS and FSC in systems that require fast rotation. The integration of TDS in state-of-the-art x-ray detectors is feasible.

Abstract 2071A: Anti-VEGF treatment in orthotopic breast cancer xenografts: Dynamic contrast-enhanced micro-CT correlates with 3D multispectral fluorescence histology

Proceedings: AACR Annual Meeting 2014; April 5-9, 2014; San Diego, CA Aim: Here, we evaluate the ability of dynamic contrast-enhanced micro-computed tomography (DCE micro-CT) in combination with 3D fluorescence ultramicroscopy (UM) to assess the early physiological and morphological treatment response of orthotopic breast tumors to anti-angiogenic drug therapy. Methods: Non-invasive DCE micro-CT was performed on breast cancer xenografts (KPL-4) that were treated twice with an antibody against the vascular endothelial growth factor (VEGF) or a control antibody. Tumor physiology was assessed in different tumor regions of interests measuring the relative blood volume (rBV) and the permeability-surface area product (PS). Parametric maps were calculated on a voxel-by-voxel basis to evaluate intra-tumor vascular heterogeneity. Mean physiological measurements were compared with morphological measures of the tumor vascular architecture (microvessel density (MVD) and relative vessel volume (rVV)) as obtained by 3-dimensional fluorescence UM. Additionally, vessel leakage was assessed by UM quantifying the penetration strength of a co-injected fluorescence-labeled therapeutic antibody into the tumor tissue. Results: Examination of the parametric maps revealed significantly different spatial patterns of intra-tumoral blood supply (rBV) between anti-VEGF-treated tumors and control. While a characteristic rim vascularization (high rBV values in the tumor periphery) was observed in the control tumors, treated tumors showed a widely homogeneous blood supply. Compared with high-resolution UM, these physiological rBV maps showed excellent agreement with the morphological appearance of the 3D tumor vascular architecture. Assessment of the mean physiology showed a significant decrease of rBV (p<0.01) and PS (p<0.05) in the tumor periphery after anti-angiogenic treatment. The reduction of rBV correlated well with a significant reduction of rVV (p<0.01) in the corresponding tumor region. The decrease in PS was found consistent with a significantly reduced antibody uptake (p<0.01) in the tumor tissue after treatment. No treatment effect was observed by tumor volume. Conclusions: DCE micro-CT used along with UM provides comprehensive and complementary information of the physiological and morphological anti-angiogenic treatment response in breast cancer xenografts. This technology may help to improve current standard methods in the assessment of anti-angiogenic drug efficacy in preclinical drug development. Citation Format: Thomas Poschinger, Anja Renner, Fabian Eisa, Michael Dobosz, Robert Brauweiler, Willi A. Kalender, Werner Scheuer. Anti-VEGF treatment in orthotopic breast cancer xenografts: Dynamic contrast-enhanced micro-CT correlates with 3D multispectral fluorescence histology. [abstract]. In: Proceedings of the 105th Annual Meeting of the American Association for Cancer Research; 2014 Apr 5-9; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2014;74(19 Suppl):Abstract nr 2071A. doi:10.1158/1538-7445.AM2014-2071A

Perfusion Measurements in Micro-CT: Increased Dose Efficiency by Use of HYPR LR Image Reconstruction

In the last decade micro-CT has become an important modality in pre-clinical research of biological processes. Micro-CT perfusion implicates long exposure times and high spatial resolution, leading to increased radiation doses in order to achieve reasonable signal-to-noise ratio (SNR). Highly constrained backprojection (HYPR) allows for dose-efficient reconstruction of serial images. A modified version of the algorithm, HYPR Local Reconstruction (HYPR LR) has been introduced recently for MR applications. We applied the HYPR LR algorithm to micro-CT perfusion data of a dedicated perfusion phantom and evaluated its performance with respect to image noise, temporal resolution and spatial resolution. In our study the image noise was reduced by 43.5% without compromising temporal resolution. Reference and HYPR LR images resulted in the same perfusion values within to 1.25%. Spatial resolution was reduced by HYPR LR by 4.75%. HYPR LR appears to be a promising method to increase dose efficiency in perfusion CT. However, further research is required to evaluate improved filter kernels and the algorithm’s dependence on spatio-temporal correlation in the dataset.