A 3-Dimensional Microvascular Phantom for Perfusion Imaging
Thomas Gaass1,2, Moritz Schneider1, Michael Ingrisch1, Julia Herzen3, and Julien Dinkel1,2

1Institute for Clinical Radiology, Ludwig-Maximilians University, Munich, Germany, 2Comprehensive Pneumology Center, German Center for Lung Research, Munich, Germany, 3Department of Physics, Technische Universit√§t M√ľnchen, Munich, Germany


The presented work demonstrates the applicability of a dedicated 3-dimensional phantom as a realistic MR- and CT-compatible phantom for microvascular perfusion simulation. The device constructed using resin-embedded, melt-spun, sacrificial sugar structures was examined using dynamic contrast enhanced MRI. Parameters, such as flow and volume fraction gained from deconvolving the signal enhancement curve showed very good agreement with the pre-set perfusion parameters. The presented phantom showed great potential in realistically simulating the capillary bed and can potentially serve as a quality insurance device for quantitative dynamic contrast enhanced MRI in the future.


Currently, only few three-dimensional phantoms are available for the accurate simulation of perfusion on a capillary level. Approaches such as stacking of 2D structures formed by lithographic methods1 or 3D printing2 either lack accuracy in at least one dimension or are too coarse for the simulation on micrometer level. Within this work we demonstrate the potential of a microvascular phantom constructed using resin-embedded, melt-spun, sacrificial sugar structures3 as a simulation tool for dynamic contrast enhanced (DCE) magnetic resonance imaging (MRI).


Phantom construction: The employed phantom, as previously introduced by Bellan et al.3, was constructed by embedding a ball of sugar fibers into a synthetic resin and subsequent dissolution of the fibers in a water ethanol bath. The sacrificial sugar structures were melt-spun using a store-bought cotton candy machine (Candyland, Klarstein, Berlin, Germany) modified in terms of rotational speed and heating temperature to control the diameter and flexibility of the fibers. A block of sugar fibers was impacted in between two plates formed from molten sugar, equipped with an in- and outlet and placed into a PET mold (cf. Fig. 1). A hydrophobic two-component resin (E45GB, Breddermann Kunstharze, Schapen, Germany) at a hardener:resin ratio of 10:6 was used to fixate the sugar structures. After approximately 24h of curing time the hardened resin block was placed in a water-ethanol bath in order for the sugar structures to dissolve, leaving a microstructure network within the resin block. DCE imaging: Controlled water flow through the phantom was induced via a Harvard Apparatus (PHD 2000, Harvard Bioscience Inc., Holliston, Massachusetts, United States). A bolus of 1ml water-gadopentetate-dimeglumine solution (Magnevist, Bayer Vital, Leverkusen, Germany) of 1% concentration followed by demineralized water was injected at a constant flow rate of 1ml/min. DCE MRI measurements were performed on a 3T whole body MRI (Siemens Skyra, Siemens Healthcare, Erlangen, Germany) using a wrist coil and a dynamically acquired TWIST sequence with the following parameters: TR/TE=4.91/1.9ms, #slices=10, slice thickness=3.5mm, FA=12°, matrix=384x384, FoV=260x260mm, temporal resolution=2.5s, TA=23min. The signal enhancement curve in the phantom was averaged over all pixels in a central slice and deconvolved with an ‘arterial input function’ measured in the supplying tube. Postprocessing was performed using regularized deconvolution with generalized cross validation4, yielding estimates of flow (F), volume fraction (v) and mean transit time (MTT). In addition to the MRI acquisition a high resolution micro-CT was acquired using a GE VtomeX M (GE Measurement and Control) with the following parameters: Voltage=70kVp, Current=120µA, Resolution=18µm.


Figure 1 depicts a 3-dimensional surface rendering (OsiriX Imaging Software) of the microvascular system within a 5x5x5mm3 cube generated from micro-CT data. Connected vessels of different diameter (4-40µm) and water filled spheres, stemming from imperfections in the fiber retrieval process are clearly visible. An intensity histogram based volumetry performed on the CT dataset yielded a mean volume fraction of the vessel network of 0.7 %. The number of enhancing voxels (voxel size: 1.604mm3) over all slices of the DCE acquisition yielded a total volume of 10.63cm3 incorporated by the ball of sugar fibers. Taking the pre-set flow rate of 1ml/min into account an average plasma flow rate of 9.4ml/100ml/min can be expected from this setting. Figure 2 displays the measured signal enhancement curve in the phantom. Deconvolution yielded F=9.9ml/100ml/min, v=0.6% and MTT=3.6s.


The visual inspection and size estimation based on the micro-CT reconstruction shows a dimensionality and structure well comparable to in vivo capillary beds5. Both the diameter and the vessel density are adjustable via the construction process and are subject of current optimizations for the specific simulation of various manifestations of capillary networks. Both the estimated flow rate of 9.4ml/100ml/min, as well as the computed network volume fraction of 0.7% are very closely reproduced by the DCE measurement results of F=9.9ml/100ml/min and v=0.6%.


The presented work demonstrates the applicability of the constructed device as a realistic MR- and CT-compatible phantom for microvascular perfusion simulation. Pre-set perfusion parameters in terms of total flow could be detected and quantified via DCE MRI in very good agreement to the true values. Future work will concentrate on a further standardization of the manufacturing process to guarantee reproducible perfusion parameters. This phantom can potentially serve as a quality insurance device for quantitative dynamic contrast enhanced MRI in the future. It may even be used as a gold standard for tracer-kinetic quantification techniques, with the limitation that the phantom currently can only provide a single, vascular compartment.


No acknowledgement found.


1) Luo Y, Zare RN, Perforated membrane method for fabricating three-dimensional polydimethylsiloxane microfluidic devices. Lab on a Chip 2008; 8:1688–1694

2) Therriault D, S. R. White SR, Lewis JA, Chaotic mixing in three-dimensional microvascular networks fabricated by direct-white assembly. Nat. Mater. 2003; 2:265–271

3) Bellan LM, Singh SP, Henderson PW, et al., Fabrication of an artificial 3-dimensional vascular network using sacrificial sugar structures. Soft Mater 2009; 5:1354-1357

4) Sourbron S, Luypaert R, Schuerbeek P, et al., Choice of the regularization parameter for perfusion quantification with MRI. Phys Med Biol 2004; 14:3307-3324

5) Freitas RA, Nanomedicine, Volume I: Basic Capabilities. Landes Bioscience, Georgetown, TX, 1999


Fig. 1: Synthetic resin-embedded sacrificial sugar structure phantom

Fig. 2: 3D surface rendering of microvasculature from micro-CT data

Fig. 3: Dynamic signal enhancement after contrast agent bolus injection

Proc. Intl. Soc. Mag. Reson. Med. 24 (2016)