EMIM 2018 ControlCenter

Online Program Overview Session: PW-08

To search for a specific ID please enter the hash sign followed by the ID number (e.g. #123).

X-Ray Imaging Technology

Session chair: Claudia Kuntner - Vienna, Austria; Christian Dullin - Göttingen, Germany
Shortcut: PW-08
Date: Thursday, 22 March, 2018, 11:30 AM
Room: Banquet Hall | level -1
Session type: Poster Session


Click on an contribution preview the abstract content.

# 082

Spectral photon-counting CT of lungs of mice infected with Mycobacterium Tuberculosis: potential for locating Tuberculosis involvement (#82)

A. Ortega-Gil1, 2, M. Moghiseh3, C. Lowe3, A. Muñoz-Barrutia1, 2, A. Raja3, A. P. H. Butler3, 4, 5, J. J. Vaquero1, 2, N. G. Anderson3

1 Universidad Carlos III de Madrdi, Departamento de Bioingenierıa e Ingenierıa Aeroespacial, Madrid, Spain
2 Instituto de Investigacion Sanitaria Gregorio Maranon (IiSGM), Madrid, Spain
3 Centre for Bioengineering, Department of Radiology, University of Otago, Christchurch, New Zealand
4 European Centre for Nuclear Research (CERN), Geneva, Switzerland
5 MARS Bioimaging Ltd, Christchurch, New Zealand


Tuberculosis (TB) infects over 9M people worldwide every year and kills 1.7M. The increase of drug-resistant strains claim for new combinations of treatments whose discovery is speed up by the use of computed tomography (CT) images [1,2]. However, distinguishing the burden and location of involvement  by TB poses a great challenge due to low contrast of lung tissue and ground glass opacity. The aim of the present research is to test the hypothesis that spectral photon-counting CT can differentiate infected and healthy tissue using cross-over contrast agent experiments in excised-mouse lungs.


Mouse organs derived from six females from the C3HeB/FeJ strain. Three were inoculated with H37Rv and sacrificed during the chronic phase of the disease. One healthy and one infected mouse lungs were stained with Silver Nitrate (AgNO3) in 3% w/v saturated solution. A further healthy and infected lungs were stained with Iodine (I2) in 3% w/v solution. One healthy and one infected lungs, unstained, served as controls. Each lung was scanned using Medipix All Resolution System (MARS) photon-counting micro-CT [3]  at clinical x-ray energy ranges (118keV, 13uA) using 18, 30, 45, and 75keV thresholds to cover the k-edges of Ag and I. Specific Ag and I linearity responses were determined from a phantom and used to measure Ag and I in Material Decomposed (MD) images of the lung samples.


The energy windows provided sufficient data to separate unstained and stained tissues by sampling the material-specific attenuation curves of unstained lung parenchyma, and lung stained with Iodine or Silver for both healthy and infected lungs. Volumetric results showed a sample’s uptake of I between 1.97 and 9.07mg/ml in the infected sample and between 2.03 and 5.39mg/ml in the healthy sample. The uptake of Ag was between 2.15 and 11.68mg/ml in the infected sample and between 2.81 and 5.46mg/ml in the healthy sample. CAs were absent in unstained samples.

Findings are discussed in terms of the concentration within lesions leading to their characterization as encapsulated cup with a peripheral rim (circular granuloma) or diffuse inflammatory response, forming a rim which lacked clearly defined boundaries.


The current study successfully showed the ability of MARS spectral photon-counting CT to differentially assess uptake of I or Ag into normal parenchyma and into TB-infected lung sites in ex-vivo whole lung. TB lesions were differentiated from normal lung tissue providing metrics for quantifying and site and burden of infective agent. In future in-vivo studies it may be possible to measure how the burden of infective agent responds to treatment. The ability to make this assessment at imaging would allow faster diagnosis and personalised treatment of difficult to diagnose lung infections.



[1] Irwin MS, Driver E, Lyon E, Schrupp C, Ryan G, Gonzalez-Juarrero M, et al. (2015) Presence of multiple lesion types with vastly different microenvironments in C3HeB/FeJ mice following aerosol infection with Mycobacterium tuberculosis. Dis Model Mech.;8(6):591–602.

[2]  . Lin L, Coleman T, Carney J, Lopresti B, Tomko J, Fillmore D, et al.(2013) Radiologic responses in cynomolgus macaques for assessing tuberculosis chemotherapy regimens. Antimicrob Agents Chemother.57(9):4237–4244.

[3] Anderson, N. G. and; Butler, A. P. (2014). Clinical applications of spectral molecular imaging: potential and challenges. Contrast media and molecular imaging, 9(1), 3-12.


The research leading to these results received funding from the Innovative Medicines Initiative (www.imi.europa.eu) Joint Undertaking under grant agreement no. 115337, whose resources comprise funding from the European Union Seventh Framework Programme (FP7/2007-2013) and EFPIA companies in kind contribution. This work was partially funded by projects RTC-2015-3772-1, TEC2015-73064-EXP and TEC2016-78052-R from the Spanish Ministry of Economy, Industry and Competitiveness (MEIC), TOPUS S2013/MIT-3024 project from the regional government of Madrid. This study (was supported by the Instituto de Salud Carlos III (Plan Estatal de I+D+i 2013-2016) and cofinanced by the European Social Fund (ESF) ‘‘ESF investing in your future’’. The authors would like to acknowledge Prof Philip Butler, Dr. Christopher Bateman, Raj K. Panta, David Knight,  Niels de Ruiter, Alex Chernoglazov and MARS members in the Centre for Bioengineering; and the Medipix2 and Medipix3 collaboration.

Attenuation and Iodine decomposed volumetric images
Lung, esophagus and trachea reconstructed scans of a healthy sample (upper row) and Tb-diseased sample (lower row). The linear attenuation images are in the engergy window from 18 to 32 KeV. This image, along with the other three energy windows, provide the information to identify I2 contrast agent (central column images) and to quantify its concentration (left column images). 
Granuloma quantification
Example of the measurements obtained for a single slice of the Iodine uptake of a lesion characterized as: A) a circular granuloma due to its organization with an encapsulated cup (brighter yellow color indicates higher concentrations of contrast agent) surrounded by a peripheral rim of Iodine lower concentration; and B) diffuse inflammatory response.
Keywords: spectral-CT, preclinical, Tuberculosis, material decomposition, photon-counting, pulmonary infection
# 083

A complete quantification study on x-ray image contrast induced by gold nanoparticles for planar imaging and Computed Tomography at preclinical and clinical energies (#258)

M. Rouchota1, 2, S. Neyt3, R. Van Holen3, G. Kagadis1, G. Loudos2, 4, 5

1 University of Patras, Department of Medical Physics, School of Medicine, Patras, Greece
2 BET Solutions, Athens, Greece
3 Molecubes, N.V., Ghent, Belgium
4 National Center for Scientific Research (NCSR) “Demokritos”, Institute of Nuclear & Radiological Sciences & Technology, Energy & Safety, Athens, Greece
5 Technological Educational Institute of Athens, Department of Biomedical Engineering, Athens, Greece


Gold nanoparticles (GNPs) are already recognized as a promising contrast agent in planar x-ray and Computed Tomography (CT) imaging, and as drug carriers for targeted therapy [1]. In both cases, being able to determine the allocation of the GNPs in vivo and defining the lowest detectable solution concentration, is of crucial importance in several applications. A complete quantification study between GNPs solution concentrations and x-ray image contrast induction has been performed and is presented for the first time.




The measurements were carried out on four x-ray imaging systems, to account for both preclinical and clinical energies (30-140 kVp). For planar imaging, a prototype benchtop system, with a spatial resolution of 0.1 mm at 30 cm source to surface distance (SSD) was used, along with a planar medical system (Siemens). For CT imaging, the commercial X-CUBE™ system (Molecubes BE), having a spatial resolution of 0.05 mm was used, along with a medical CT system (Lightspeed GE). The image contrast was quantified through Contrast to Noise Ratio (CNR) [2] for planar imaging and through Hounsfield Units for CT imaging. GNPs samples (AuroVist™ 15 nm Nanoprobes) of 0.01-200 mg GNPs/ml were studied and compared to standard iodine solutions. Selected concentrations were validated through in vivo imaging.



The imaging studies showed that for preclinical planar imaging, where an x-ray tube and a CMOS detector were used at an SSD of 30 cm, concentrations as low as 2 mg Au/ml were visible, i.e. presenting a CNR > 5. The CNR slightly increased with higher energies. For the clinical planar x-ray system, higher concentrations were required and solutions started being noticeable at around 10 mg Au/ml. CT imaging greatly improved the detection sensitivity, as expected [1]. For the clinical CT system, solutions of 0.1 mg Au/ml induced contrast values around 7 HUs and became clearly evident at 2 - 5 mgAu/ml, presenting contrast values of 70 - 150 HUs. The lowest detectable concentration was found through the X-CUBE™ system, before post-processing application. When the Dual Energy imaging method was applied, even lower concentrations were detectable. The lowest detectable concentrations were validated through in vivo imaging and a full set of in vivo measurements was acquired for selected samples.



A full quantification study between GNPs' (15 nm) concentration and x-ray image contrast induction has been performed, over a wide range of solutions and energies, both at preclinical and clinical level. The minimum detectable concentration has been defined for each mode and selected indicative concentrations have also been validated through in vivo imaging. The strength of Dual Energy imaging has also been demonstrated on very low concentrations, not detectable through the standard imaging protocols.




  1. Hainfeld, J. F., Slatkin, D. N., Focella, T. M., & Smilowitz, H. M. (2006). Gold nanoparticles: a new X-ray contrast agent. The British journal of radiology, 79(939), 248-253.
  2. Mori, Masaki, Kuniharu Imai, Mitsuru Ikeda, Youko Iida, Fukiko Ito, Kazuo Yoneda, and Yukihiro Enchi. "Method of measuring contrast‐to‐noise ratio (CNR) in non uniform image area in digital radiography." Electronics and Communications in Japan 96, no. 7 (2013): 32-41.




This work was supported by the Research Projects for Excellence IKY/SIEMENS and the results have been used as preliminary results for the European Commission H2020-NMBP-2017 funded n-TRACK project (761031-2). The authors would like to acknowledge the AlphaVet clinic for providing the clinical systems used in this study.


Figure 1
Quantification between GNPs concentration and x-ray contrast was based on phantom imaging studies. An example for the preclinical planar x-ray and four different energies is shown.
Figure 2
Pre and post contrast CT images of a mouse that was injected with 200 µL of 50 mg/mL solution, acquired at 40 kV.


Keywords: Gold Nanoparticles, x-ray imaging, Computed Tomography
# 084

Combined mouse brain and whole body skeleton imaging within a single µCT scan (#372)

S. Llambrich Ferré1, J. Wouters1, N. Martínez-Abadías2, 3, 4, G. Vande Velde1

1 KU Leuven, Biomedical MRI, Department of Imaging and Pathology, Leuven, Belgium
2 Center for Genomic Regulation (CRG), The Barcelona Institute of Science and Technology, Barcelona, Spain
3 Universitat Pompeu Fabra (UPF), Barcelona, Spain
4 European Molecular Biology Lab (EMBL), Barcelona, Spain


Imaging of both the skull and the brain is currently done via co-registration of data from CT and MRI scans. This is a time and cost-intensive workflow that requires multiple acquisition and subsequent co-registration steps. Our aim is to develop a technique that allows the visualization of both systems in a single acquisition, thereby reducing the costs and time and obviating the need for offline co-registration. We optimized diceCT1, a technique that uses iodine as a contrast agent to visualize soft tissues in a CT scan, to enable integrated and accurate mouse brain and bone quantification.


Whole-body micro-CT-scans with a reconstructed voxel size of 50 µm isotropic were acquired on a Skyscan 1278 microCT scanner. We tested three different staining approaches using Lugol as the contrast agent: perfusion, diffusion and a novel stereotactic injection approach. With perfusion we tested different injection rates, concentrations of Lugol and paraformaldehyde (PFA). Regarding the diffusion approach we optimized the protocol considering the freshness of tissue, fixation time, incubation time in Lugol and frequency of changing to fresh Lugol. Finally, as for the stereotactic injection in the brain ventricles, we optimized the volume and injection rate of Lugol, and injection rates in one or the two lateral ventricles, before and after fixation.


Via trans-cardiac perfusion in combination with blood-brain barrier disruption, we could not reach homogeneous contrast enhancement in the brain. For the diffusion approach, we optimized a protocol to obtain highest brain contrast with minimal shrinkage, allowing us to clearly analyze the brain volume and visualize different brain regions. Regarding our novel stereotactic injection approach, we have been able to clearly visualize the brain ventricles, allowing ventricle segmentation and volume analysis. The skull was visible in all the approaches.


Although the co-registration of MRI and CT is currently the gold standard and there is still room for optimization of our current approach, we here propose two different novel approaches to visualize both the skull and soft tissue of the brain in a single scan, obviating the need for multimodal acquisition and co-registration. Our approach is readily applicable for mouse model evaluation in Neuroscience and Developmental Biology.


1.        Gignac, P. M. et al. Diffusible iodine-based contrast-enhanced computed tomography (diceCT): an emerging tool for rapid, high-resolution, 3-D imaging of metazoan soft tissues. J. Anat. 228, 889–909 (2016).

Keywords: DiceCT, Skull & Brain Imaging, Contrast CT
# 085

High resolution visualization and quantification of murine renal fibrosis by micro computed tomography (#341)

J. Missbach-Guentner1, D. Pinkert-Leetsch1, C. Dullin1, 2, D. Hornung3, B. Tampe4, M. Zeisberg4, F. Alves1, 5, 6

1 University Medical Center Goettingen, Institute of Diagnostic and Interventional Radiology, Goettingen, Lower Saxony, Germany
2 Italian Synchrotron Light Source "Elettra", SYRMEP beamline, Trieste, Italy
3 Max Planck Institute for Dynamics and Self-Organisation, Group of Biomedical Physics, Goettingen, Lower Saxony, Germany
4 University Medical Center Goettingen, Department of Nephrology and Rheumatology, Goettingen, Lower Saxony, Germany
5 University Medical Center Goettingen, Clinic of Hematology and Medical Oncology, Goettingen, Lower Saxony, Germany
6 Max Planck Institute for Experimental Medicine, Translational Molecular Imaging, Goettingen, Lower Saxony, Germany


The increasing number of patients with end stage chronic kidney disease not only calls for novel therapeutics but also for pioneering research using convincing preclinical disease models and innovative analytical techniques.

The aim of this study was to introduce a virtual histology approach using contrast enhanced micro computed tomography (µCT) for the entire murine kidney. Generating a three dimensional (3D) high resolution dataset allowed not only the visualization but also the quantification of renal fibrotic remodeling based on altered radio-opacity in a ureteral obstruction model.


Four FVB/N mice underwent surgery for a unilateral ureteral obstruction (UUO) procedure as described earlier by Tampe et al. 2015. After 10 days, the ureteral ligation was removed and the animals recovered for another 7 days before sacrifice and kidney extraction.

For staining and fixation, the excised kidneys were placed in a staining solution containing phosphotungstic acid (PTA). The PTA staining process was inspected by CT scans and ranged from 4 to 10 days to complete penetration of the kidneys.

The kidneys (fixed only or embedded in agarose) were scanned either with the QuantumFX in vivo µCT (Perkin Elmer) or with the Nanotom specimen CT (Phoenix, GE). Data reconstruction was performed automatically directly after the scan by the CT-specific software.


Ex vivo PTA staining of the UUO and corresponding sham kidneys not only allowed the visualization of the 3D entire kidney morphology but enabled the quantification of certain renal morphological structures due to the distinct uptake of PTA. Using the software SCRY, a histogram was generated from the µCT data sets, representing the different distribution of x-ray absorption throughout the kidney, displayed by grey values (GV). The GV definition and distribution enabled the quantification of the kidney volume, the volume of the single cortex, medulla and pelvis and the distinct cortical thickness as surrogate marker for the aging/ pathologically altered kidney. Furthermore, the altered uptake of PTA in UUO vs sham kidneys revealed a significant reduction of radio opacity within the renal cortex, loss of cortical mass and a loss of cortical thickness. The detected fibrotic tissue remodeling was confirmed by histological and IHC analysis, which was not impeded by the PTA staining.


Our results show that PTA enhanced µCT analysis in preclinical studies can help to understand mechanisms of kidney failure by revealing and quantifying pathological alterations in renal diseases and thus support the optimization of therapeutic strategies and refine the outcome of preclinical research on kidney associated murine disease models. Furthermore, this method can be easily adapted to other murine organs (Dullin et al. 2017).


Tampe, B., Tampe, D., Zeisberg, E.M., Müller, G.A., Bechtel-Walz, W. et al. Induction of Tet3-dependent epigenetic remodeling by low-dose hydralazine attenuates progression of chronic kidney disease. EBioMedicine 2, 19-36 (2015).

Dullin, C., Ufartes, R., Larsson, E., Martin, S., Lazzarini, M. et al. µCT of ex-vivo stained mouse hearts and embryos enables a precise match between 3D virtual histology, classical histology and immunohistochemistry. PLOS ONE doi: 10,1371/journal.pone.0170597 (2017).


We thank Julia Schirmer and Bärbel Heidrich for excellent technical assistance in animal work, staining and embedding procedures. We thank Roswitha Streich, Bettina Jeep, Sabine Wolfgramm and Sarah Rinkleff for performing histology and immunostaining of tissue sections and Christina Malowsky for CT scans of isolated kidneys.

Keywords: virtual histology by micro computed tomography, ex vivo staining, quantitative analysis, renal fibrosis, 3D kidney morphology
# 086

Tumour Volume Measurement. Reliability and Accuracy of Calliper vs uCT in a Xenograph Breast Cancer Murine Model (#44)

J. A. Camara1, A. Pujol1, A. Martín-Pardillos2, 3, S. Ramon y Cajal2, 3

1 Vall d´Hebron Institute of Research, Preclinical Imaging Platform, Barcelona, Barcelona, Spain
2 Vall d´Hebron Institute of Research, Translational Molecular Pathology, Barcelona, Barcelona, Spain
3 Centro de Investigacion Biomedica en Red de Cancer (CIBERONC), Barcelona, Spain


Animal models in cancer research play an increasing key role in modern medicine, due to xenograph models that lead to an almost individualized medicine and treatment protocols. Evaluation of tumour growth is one of the most used tool for evaluating treatments effect.

Imaging techniques are widely used for monitoring internal tumour growth due to their repeatability and almost harmlessness.

We make a comparison in subcutaneous tumours between calliper and uCT measurements, to evaluate the benefits of this imaging technique in external tumour volume evaluation.


MDA-MB231 tumour cells were implanted into intramammary fat pad (IFP) in 15 athymic nude-FOXn1. Tumour volume was measured at 12, 19 and 26 days.

Calliper measurements of long and short axis followed by uCT scan were obtained in anesthetized animals. (Quantum FX. Perkin Elmer. EEUU). Acquisition parameters were: FOV 20 mm, 120 sec, 90 kV and 160 mA. Reconstruction was performed based on Feldkamp´s method. Measurement of three axis was done in uCT studies and tumour volume was calculated applying ellipsoid theoretical volume formula in both techniques.

For Gold standard, a manual “slice by slice” contour delimitation was performed in uCT image and volumetrical value was obtained. Statistical analysis was done by a Pearson product moment correlation. Figure 1


Tumour growth and volume were confirmed with both techniques. Numeric values of every animal and time point are displayed in figure 2.

There is no significant difference between values obtained with calliper or uCT techniques and gold standard (p<0.05). On the other hand, accuracy of uCT technique is higher.

Oposite to Jensen and Brodin, our results fit perfectly with gold standard measures in both techniques. Differences with their results could be related to measure of gold standard volumes. In their case, a density formula was used, in an indirect way to measure tumour volume. Our gold standard value was a real volume measurement, without removing the tumors.

The protocol for measuring uCT tumor volume is different too. Our is close to calliper way, while they used semiautomatic protocols of measure, including tumour densities and tresholds.

Animal models were different too, as we made IFP tumours instead of flank or limb placements.


Both volume values and growth rates have no statistical significance in calliper and uCT.In our opinion, calliper measures are perfectly accurate for IFP models, less complicated and cheaper than uCT scans. Further studies should be developed to find the reason of our contradictory results with published data.


Jensen M, Jørgensen JT, Binderup T, and Kjærcorresponding A. Tumor volume in subcutaneous mouse xenografts measured by microCT is more accurate and reproducible than determined by 18F-FDG-microPET or external caliper. BMC Med Imaging. 2008; 8: 16.

Brodin NP, Tang J,Skalina K, Quinn TJ, Basu I, Guha C, and Tomé WA. Semi-automatic cone beam CT segmentation of in vivo pre-clinical subcutaneous tumours provides an efficient non-invasive alternative for tumour volume measurements. Br J Radiol. June 2015; 88.


We will like to thank Functional Validation and Preclinical Research group and Laboratory Animal Service at Vall d´Hebron Research Institute for their work and help during this experiment.

The authors declare no competing interests.

Figure 1
Left: uCT measurement of tumor in three diferent axis. Right: 3D rendering of tumour isocontour after "slice by slice" segmentation
Figure 2
Left: numeric values of tumours including real, uCT and calliper.  Right: different measurement correlations in different days (up) and whithout time consideration (middle). Values of growth rates are correlated at bottom.
Keywords: volume, CT, calliper, xenograft, tumour
# 087

Longitudinal Imaging of an Intravenous Lung Tumour Murine Model. Comparison of Optical Imaging and uCT Techniques (#48)

J. A. Camara1, A. Pujol1

1 Vall d´Hebron Institute of Research, Preclinical Imaging Platform, Barcelona, Barcelona, Spain


The use of animal models in cancer research plays a relevant role in modern medicine. The possibility to reproduce tumour progression in a real living environment puts this modality over in vitro studies.

Imaging techniques are widely used for monitoring tumour growth. Their repeatability and almost harmless processes give the possibility to make longitudinal studies in most of the experiments. Optical imaging (bioluminescence modality) and uCT are the most used imaging techniques in oncology.

This work makes a comparison between both techniques in an intravenously lung tumour mouse model.


Eight NRMI female were injected intravenously via tail vein with KLN205 cancer cells in a dosage of 300.000 cells for six animals and 150.000 cells for the other two. At days 3,10,14 and 28, optical imaging bioluminescence (BLI) and uCT acquisitions were obtained. For BLI, 150mg/kg of luciferine were injected intraperitoneally, followed by a 20-25 minutes kinetic acquisition of images. uCT scans were performed with 24mm field of view, 2 minutes acquisition and 90kV/160uA energy.

In every BLI image, ROI´s were generated including the whole chest and background. Maximum signal time point was searched in every subject and images were analysed for positive signals. uCT images were analysed, and size and radiological density of every tumour was measured. Healthy lung density was measured. Fig 1


Regarding uCT technique, sensibility was 1,48mm3 tumour size, and 0,13 signal/noise ratio (SNR). All the animals were positive from the beginning and a clear evolution of tumour number, size and radiological density was confirmed. Two different behaviours were observed in tumour evolution. First, one with less changes in tumour volume and density, followed by a fast development of tumours, including quick growth and increase of radiological density.

BLI images were positive in almost the images (3 negatives in first time point) and lowest positive SNR was 15. Increasing values of light were observed. Like in uCT, there was a different evolution of light emission. A flat period during the first days leads to an acute increase of the signal values at the end of the experiment.

Animals with lower cell doses presented differences, with lower signals in BLI, and low tumour volume in uCT at the end of the experiment, but not at the beginning. Radiological density didn´t vary. Fig 2


Data show a similar behaviour in results of both techniques. A flat stage with low light emission and small tumours and radiodensity is followed by an increasing of light emission as well as tumour volume and harder masses.

There is no association between light, volume and density during different time points in flat stage, while there are related during the second phase of the tumour development. Looking at the data in a non-time way, there is a strong relation between light intensity, tumour volume and radiological density.

Both techniques could be suitable for lung tumour evaluation.


Patel P, Kato T, Ujiie H, Wada H, Lee D, Hu HP, Hirohashi K, Ahn JY, Zheng J, Yasufuku K. Multi-Modal Imaging in a Mouse Model of Orthotopic Lung Cancer. PLoS One. 2016 Sep 1;11(9)

Davison CA, Chapman SE, Sasser TA, Wathen C, Diener J, Schafer ZT, Leevy WM. Multimodal optical, X-ray CT, and SPECT imaging of a mouse model of breast cancer lung metastasis. Curr Mol Med. 2013 Mar;13(3):368-76.


We will like to thank Laboratory Animal Service at Vall d´Hebron Research Institute for helping with animal management.

Figure 1
Left: uCT images with a highligthed tumour (A) and complete lung render (B). Right: analysis of BLI study (C), application of same ROI´s in a sequence of images (D) and examples of two kinetic curves of bioluminiscence
Figure 2
Left: time line of average radiance in BLI studies (A) and uCT tumour volume and density (B). Both techniques present same temporal behaviour. Right: correlation between BLI signal/noise ratio and tumour volume (C) and density (D). Both are positive and statistically significant.
Keywords: tumour, animal, optical imaging, ct, lung
# 088

µCT imaging approach for mouse lung fibrosis detection induced by bleomycin osmotic mini pump (#558)

F. Ruscitti1, S. Belenkov2, F. Ravanetti3, V. Bertani3, R. Ciccimarra3, V. Menozzi3, J. Essers4, Y. Ridwan4, A. KleinJan5, P. Van Heijningen4, A. Cacchioli3, M. Civelli1, G. Villetti1, F. F. Stellari1

1 Chiesi S.p.A., PreClinical R&D, Parma, Parma, Italy
2 PerkinElmer Inc., Waltham, Massachusetts, United States of America
3 University of Parma, Veterinary Department, Parma, Parma, Italy
4 Erasmus MC, Department of Molecular Genetics, Vascular Surgery, and Radiation Oncology, Rotterdam, Netherlands
5 Erasmus MC, Department of Pulmonary Medicine, Rotterdam, Netherlands


Bleomycin (BLM) subcutaneous infusion by osmotic minipump has been reported to induce lung  fibrosis in mice that more closely resembles human interstitial lung disease compared to direct delivery into the lungs1.  In this study, staging of pulmonary fibrosis and efficacy of Nintedanib treatment (FDA approved drug) have been evaluated by correlating histological parameters of fibrotic lesions to different degrees of aeration in lung compartments measured by in vivo mCT.


Osmotic minipumps (delivery rate 0.5 µl/h) containing either vehicle or BLM (100 mg/kg) were implanted subcutaneously for 7 days. Three groups infused with BLM were treated daily with vehicle and Nintedanib at 30-60 mg/Kg o.s. for 14 days starting at the end of the second week. At day 28 mCT scans were acquired using Quantum GX mCT (PerkinElmer, Inc.) and lung volumes were calculated with semi-automatic segmentation using Perkin Elmer Analyzer software. For quantitative assessment, Hounsfield Unit (HU) clinical ranges were applied on rescaled HU images and lung tissue resulted in a division between normally and poorly aerated regions2. Finally, lungs were isolated for histology, and Ashcroft score was quantified and compared to mCT parameters.


mCT analysis detected a significantly increase of poorly aerated lung tissue percentage in BLM (88±8.2) compared to vehicle group (19.7±2.5). Nintedanib at 60 mg/kg reduced the percentage of poorly aerated tissue (42.6±10.3; p<0.05), while no effect of treatment was observed at 30 mg/kg. Lung histology revealed the presence of isolated fibrotic foci located in subpleural region, while the major changes are observed in the parenchyma characterized by diffuse fibrosis and alveolar septa thickening  in BLM compared to vehicle group. Significant inhibition of the Ashcroft score (39%; p<0,01) has been detected only in Nintedanib treated group at the higher dose in agreement with mCT data


The most relevant information from µCT analysis is that the increased poorly aerated lung tissue observed in BLM-pump model is consistent with fibrotic lesions detected by classical histology. µCT can reliably give 3D information on lung disease changes and it could be considered useful tool either to understand the basis of fibrosis pathology or to evaluate efficacy of new anti-fibrotic drugs.


  1. Lee R. et al. Bleomycin delivery by osmotic minipump: similarity to human scleroderma interstitial lung disease Am J Physiol Lung Cell Mol Physiol 2014 Apr 15;306(8):L736-48.
  2. Gattinoni L. et al. What has Computed Tomography taught us about the acute respiratory distress syndrome? Am J Respir Crit Care Med 2001 Nov 1;164(9):1701-11.
Tissue degrees of aeration in BLM PUMP mouse model
Percentage of poorly and normally aerated tissue are presented in Figure 1 for Vehicle,Bleomycin and Nintedanib treated groups. On the left, representative images of the segmentation object map form uCT 
Correlation uCT and histology parameter
The plot shows a good correlation between the poorly-aerated tissue percentage and aschroft score
Keywords: Lung fibrosis, Micro-CT, imaging, animal model, anti-fibrotic drug
# 089

Calibration set-up for Dual Energy capabilities in a Real Advance Digital Radiography System (#349)

C. de Molina1, 2, C. Martínez1, 2, M. Desco Menéndez1, 2, 3, M. Abella Garcia1, 2, 3

1 Universidad Carlos III de Madrid, Dept. Bioingeniería e Ingeniería Aeroespacial, Madrid, Spain
2 Instituto de Investigación Sanitaria Gregorio Marañón, Madrid, Spain
3 Centro Nacional de Investigaciones Cardiovasculares Carlos III (CNIC), Madrid, Spain


Dual energy (DE) radiography, based on the acquisition of two images at different voltages, enables the separation of soft tissue and bony structures.

One approach widely used to extract the soft tissue and bone maps is based on the characterization of the attenuation of x-rays by acquiring a phantom made of soft tissue and bone equivalents at two energies. Most works [1,2] are based on simulations but do not address practical issues of the implementation  in a real system.

We present a complete calibration protocol to incorporate DE capabilities in an X-ray system including the phantom design.


We studied the optimum design of the DE phantom (material and size) using simulations, evaluating the attenuation characterization and the resulting tissue maps. Regarding the equivalent materials, we studied water, PMMA and plastic A-150 as soft-tissue equivalents and Teflon, plastic B-100 and aluminum as bone substitutes, considering their physical density, mean ratio of Z/A, X-ray mass attenuation coefficients (MAC), manufacturability and cost.

To obtain the optimum size, we simulated wedge phantoms with sizes of 50, 100, 150, 200 and 250 mm and a voxel size of 0.07 mm. Evaluation of the proposed calibration protocol was done on the NOVA FA X-ray (SEDECAL) with a thorax phantom.


PMMA was chosen as soft-tissue equivalent material since it presents small difference in terms of Z/A ratio and MAC (0.01 and 0.08 cm2/gr respectively) and it allows an easy manufacture of a solid phantom with a low cost (around 50 €). For cortical bone substitute, aluminum is the most suitable material but since it is non-machinable in its pure state, we have chosen the alloy Al 6082. The phantom with a size side of 100 mm resulted large enough to cover the traversed thicknesses of any anatomical study.

The calibration protocol has four steps: (1) Laser-guided positioning of the phantom; (2) 170/40 KVp acquisitions; (3) Estimation of traversed thicknesses; (4) Registration of acquired data and estimated thicknesses; (5) Model fitting of the DE characterization.

Results with the PBU-60 phantom are shown in Figure 1.


We have presented a complete protocol to incorporate DE capabilities in a radiography system that does not need the estimation of the spectra or the mass attenuation coefficients.

We studied the technical considerations, manufacturability and cost of the calibration phantom and selected a feasible, effective and low cost calibration phantom that imitates the soft tissue and bone properties and allows a good modeling of the DE characterization of any system.

Tests on a real system show the feasibility of the proposal with a good tissue-map extraction.


[1] Lehmann LA, Medical Physics, 1981. 8(5): p. 659-667.

[2] Cardinal HN1, Medical Physics, 1990. 17(3): p. 327-341.


This work is partially funded by the project DPI-2016-79075-R from the Spanish Ministerio de Economía, Industria y Competitividad.

Figure 1 Dual-energy decomposition using a real system

Top: Low, high energy radiographies of the thorax of the antropomorphic phantom PBU-60

Bottom: Soft-tissue and bone maps extracted using the proposed calibration protocol

Keywords: Dual Energy Radiography, Dual-energy calibration, Dual-energy substraction, Dual-energy decomposition
# 090

Recovering density values on small animal X-ray imaging through beam hardening compensation: preliminary results (#280)

C. Martínez1, 2, N. Ballesteros1, 2, C. de Molina1, 2, M. Desco Menéndez1, 2, 3, M. Abella Garcia1, 2, 3

1 Universidad Carlos III de Madrid, Departamento de BIoingeniería e Ingeniería aeroespacial, Madrid, Spain
2 Instituto de Investigación Sanitaria Gregorio Marañón, Madrid, Spain
3 Centro Nacional de Investigaciones Cardiovasculares Carlos III, Madrid, Spain


Due to the polychromatic nature of the spectra and the energy dependency of the tissue attenuation, beam mean energy is increased as it traverses the sample. Beam-hardening effect eliminates the linear relation between the traversed tissue and the measured attenuation reconstructing a map of tissue densities incorrect. It also results in two artifacts: cupping and streaks. Most scanners include a method to restore the linear relationship between measured attenuation and traversed soft tissue, based on a calibration step with a soft-tissue equivalent phantom and correcting only cupping [1].


Small animal tissues can be roughly grouped in soft tissue and bone attending to attenuation properties. However, classic linearization only takes into account the soft tissue to restore the linear relationship. Proposed method restores the linear relationship using soft tissue and bone to recover the true density values.

To find the linearization function, we need to characterize the relationship between the measured attenuation and different combinations of traversed soft tissue and bone. This could be done with a phantom composed of soft tissue and bone equivalents, but it is difficult to find equivalent materials. Proposed method eliminates materials search, finding the correction parameters intrinsically using the information of the sample itself. The workflow is shown in Figure 1.


Evaluation was done with simulations of a 3D atlas of a mouse and five rodent studies acquired with the ARGUS-CT scanner. Results showed 97% and 99% of density values restoration for soft tissue and bone respectively. The figure 2 shows the streak artefact removal achieved by the proposed method on an axial slice of a rat study for the real cases.


We have proposed a simple method to compensate the beam hardening effect in CT, reducing for both characteristic artifacts (cupping and streaks) and recovering the true density values.


[1]. R. A. Brooks and G. D. Chiro, "Beam hardening in x-ray reconstruction tomography," Phys. Med. Biol., vol. 21, pp. 390-8, 1976.


This work is partially funded by the project DPI-2016-79075-R from the Spanish Ministerio de Economía, Industria y Competitividad.

Fig 1: Workflow of the proposed method
Fig 2. Axial slice of a rat study before (left) and after (right) the correction.
Keywords: Beam-hardening, CT, Linearization, Artifact
# 091

X-Ray scatter correction in 2D radiography using a beam-stopper: a simulation study (#272)

A. Martinez Martinez1, C. Martinez Sanchez2, M. Desco Menéndez1, 2, 3, M. Abella Garcia2, 3

1 Instituto de Investigación Sanitaria Gregorio Marañón, Madrid, Spain
2 Universidad Carlos III de Madrid, Bioengineering And Aerospacial Engineering, Madrid, Spain
3 Centro Nacional de Investigaciones Cardiovasculares Carlos III, Madrid, Spain


X-Ray scatter causes a decrease in image contrast in planar radiography that can reduce its diagnostic utility. The use of anti-scatter grids allows the partial removal of the scattered photons reaching the detector at the cost of a 3-6 times increase in radiation dose. Digital radiography allows the use of post-processing methods to correct the effects of scatter, avoiding the use of anti-scatter grids. Most proposed methods in the literature focus on computer tomography, combining scatter estimation and iterative reconstruction algorithms.


We present a method for scatter correction in planar radiography based on the estimation of the scatter field from a second acquisition with a beam stopper placed between the X-Ray source and the patient. The beam stopper is a thin plate of highly absorbent material with a rectangular matrix of holes that allow the passing of the x-ray beam. The scattered radiation can be sampled in the shadowed zones of the projection image and used to estimate the scatter field by means of two-dimensional interpolation. Finally, a weighted version of the estimated scatter field is subtracted from the projection image to obtain the corrected image.

We have performed a preliminary evaluation of the method using the MC-GPU package [1] to simulate a thorax acquisition.


Figure 1 shows the scatter field estimation using a spectrum of 110kVp generated with the Spektr toolkit [2], a source-to-detector distance of 1500 mm, and a beam stopper designed to produce a grid of holes in the projection image with a diameter of 120 mm and a separation between centers of 240 mm. A total number of 270 samples were taken from the shadowed zone to estimate the scatter field.

Figure 2 represents the results of the correction with the estimated scatter field. The corrected image shows overall contrast enhancement. The plot represents a central profile from the non-corrected image, the corrected image and the scatter-free primary image. The RMSE decrease in the corrected image with respect to the non-corrected image is 80.2 % for the horizontal profile and 79.5% for the vertical profile.


Preliminary results show a decrease of around 80% in the RMSE of the corrected image with respect to the non-corrected image. The acquisition of a second image increases the radiation dose only about 20%, which compared to an acquisition with anti-scatter grids with grid factors of 3 and 6, would imply a dose reduction of 60% and 80% respectively. Further research will also focus on determining the optimal quantity and spatial distribution of the scatter samples to enable its application in flexible acquisition configurations, such as tomosynthesis protocols.


[1] A. Sisniega, M. Abella, E. Lage, M. Desco, and J. J. Vaquero, "Automatic Monte-Carlo Based Scatter Correction For X-ray cone-beam CT using general purpose graphic processing units (GP-GPU): A feasibility study," in 2011 IEEE Nuclear Science Symposium and Medical Imaging Conference (NSS/MIC), 2011, pp. 3705-3709.

[2] J. H. Siewerdsen, A. M. Waese, D. J. Moseley, S. Richard, and D. A. Jaffray, "Spektr: A computational tool for x-ray spectral analysis and imaging system optimization," Medical Physics, vol. 31, pp. 3057-3067, 2004.


This work is funded by DPI2016-79075-R from the Ministerio de Economía, Industria y Competitividad (www.mineco.gob.es/).

Figure 1 - Scatter field estimation
Beam-stopper projection (1). Sampling positions on the beam stopper projection with display range changed to enhance the visualization of the scatter signal present in the shadowed regions (2). Estimated scatter field (3). Simulated scatter field (4).
Figure 2 - Scatter correction evaluation

Normal projection (left), corrected projection (center), profile comparison (right).

Keywords: X-Ray, Scatter, Radiography, Antiscatter grid
# 092

Surface extraction with structured-light scanner for limited angle tomography (#291)

N. Ballesteros1, 2, C. de Molina1, 2, X. Ye1, M. Desco Menéndez1, 2, 3, M. Abella Garcia1, 2, 3

1 Universidad Carlos III de Madrid, Bioingeniería e Ingeniería Aeroespacial, Madrid, Spain
2 Instituto de Investigación Sanitaria Gregorio Marañón, Madrid, Spain
3 Centro Nacional de Investigaciones Cardiovasculares Carlos III, Madrid, Spain


In limited-view CT studies only few projections can be obtained within a reduced angular range, resulting in artifacts in the reconstructed images when conventional reconstruction methods are used. Advanced reconstruction methods can compensate the lack of projections by including prior information. The surface of the sample has been shown to be a good prior in  this context by constricting the solution to the subspace comprised by the surface. We propose a complete protocol using a structured-light 3D scanner to extract the surface of the sample in order to solve the reconstruction problem.


Figure 1 shows the proposed workflow.

Three non-metallic radiopaque markers are placed on the skin of the animal. We acquire a complete CT study of the rodent with 360 projections. To simulate limited data, we select 120 projections over an angular range of 120º.  

We use the scanner 3D Artec Eva to scan the surface of the animal, and the software Artec Studio to create a polygonal mesh from which a 3D binary mask of the surface is obtained.

A preliminary CT reconstruction is generated based on the FDK algorithm using Mangoose [1]. A rigid transformation is applied to the mask using the markers as fiducial points to orientate and adjust it to the field of view of the CT volume. The registered mask is then used as a prior in the advanced reconstruction described in [2].


Figure 2.A shows the ideal mask, obtained by thresholding of the preliminary CT, and the mask obtained from the surface scanner. The Sorensen-Dice index between them showed a 90.7% of similarity. The main difference is in the hair of the animal, seen by the light scanner but invisible for the CT.

Figure 2.B and 2.C show an axial slice reconstructed with an FDK-based method for full data (360 projections over 360 degrees) and limited data (120 projections over 120 degrees) respectively. In the second image, we can see the streaks and the edge distortion as a consequence of the lack of projections and the limited angular range. Figure 2.D and 2.E show the result obtained for the limited-data scenario with the iterative advanced method using the ideal and the surface masks respectively. In both cases, the contour of the sample is properly recovered, reducing the streaks artifacts, with an increase of error of only 2.44% when using the mask from the scanner.


We have presented a complete protocol to obtain tomographic images in scenarios of limited-angle data using an advanced reconstruction algorithm that includes prior information about the surface of the sample. The results on small-animal data showed the viability of a structured-light scanner for the extraction of the surface to serve as prior information. The use of this technology allows to scan the sample easily in a short period of time, avoiding the need for any additional hardware installation or contact with the sample.


[1]  Abella, M., J.J. Vaquero, A. Sisniega, J. Pascau, A. Udías, V. García, I. Vidal, and M. Desco, Software Architecture for Multi-Bed FDK-based Reconstruction in Xray CT Scanners. Computer methods and programs in biomedicine, 2011. (in press). 685618.

[2] Molina C de, Abascal JFPJ, Desco M, Abella M. Study of the possibilities of Surface-Constrained Compressed Sensing (SCCS) Method for Limited-View Tomography in CBCT systems. Proceedings of 4th International Conference on Image Formation in X-Ray Computed Tomography, 2016


This work has been funded by Ministerio de Economía, Industria y Competitividad (projects DPI2016-79075-R).

Figure 1

Proposed workflow for the Surface Constrained Method for Limited Data Tomography

Figure 2
A. Ideal mask from CT segmentation (white) over the mask obtained from the surface scanner (gray). B. FDK reconstruction of a full data (360 projections over 360 degrees). C. FDK reconstruction of limited-angle data. D. Advanced reconstruction of limited-angle data using the ideal surface. E. Advanced reconstruction of limited-angle data using the mask generated with the surface scanner.
Keywords: Computed Tomography, Limited-angle