5037

A 3D-Printed Physiologically-Faithful Perfusion Phantom that Recapitulates Microvasculature Structure for Quantitative Experimental Validation of Fluid Transport

John Morgan¹, Thanh D. Nguyen², Pascal Spincemaille³, and Yi Wang⁴

¹Cornell University, Cornell Medical College, New York, NY, United States, ²Cornell Medical College, New York, NY, United States, ³Cornell University Medical College, New York, NY, United States, ⁴Cornell University, Cornell University Medical College, New York, NY, United States

Synopsis

Recapitulation of microvascular structure, function and perfusion in vitro can enable studies of vascular biology, provide a model for diseases such as ischemic stroke or tumor angiogenesis and enable quantitative evaluation of physiologic blood or lymph perfusion. Here we describe the initial design and deployment of a first-generation, self-contained 3D-printed, physiologically-faithful, microfluidic perfusion phantom to form explicit, hierarchically-branching, microvascular structure encapsulated in a type I collagen matrix in vitro, with pump-driven perfusion easily visible via phase-contrast MRI (Fig. 1). The phantom flexibly supports creation of user-defined vessel network geometries with human vascular cells and allows experimental validation of blood flow, i.e., via constitutive equations for convective and diffusive transport that quantitatively relate the flux of tracers from time-resolved images to transport field quantities. Thus, the largely qualitative and unmeasurable global arterial input assumption in the traditional Kety’s method can be replaced with measurable and reproducible MRI experimental data, formulated as quantitative transport mapping (QTM). Preliminary data demonstrate that the QTM phantom is promising for characterizing actual blood transport in vitro in healthy and pathological contexts.

Introduction

Methods

The microfluidic perfusion phantom is entirely 3D printed, fully MRI compatible, and easily assembled, with inlet/outlet pump connections to perfuse fluid through a hermetically sealed chamber encapsulating microvasculature networks (Fig 2). It was 3D-printed using a Stratasys BJE 260 Connex 3D printer, from stereolithographic files created in AutoCAD™ software. A (poly)-dimethylsiloxane (PDMS) stamp was cast from a 3D-printed mold, containing a network channel template housed in a casting jig (Fig 3). The device was assembled as described previously.1. The top piece was placed atop the PDMS stamp, containing the (raised-feature) channel network. Type-I collagen was injected, forming around the network and bottom piece covered with a flat later of collagen. After curing, the pieces were joined hermetically using a medical-grade silicon O-ring, secured with 3D-printed screws. The microvessels were seeded with vascular cells expressing green fluorescent protein and cultured in an incubator under physiologic perfusion (shear stress ~1.5Pa). The vascular network was imaged via fluorescent microcopy before MRI data acquisition. The device was connected to a pulsatile pump and perfused at physiologic rates. A pulse of gadolinium was injected as a tracer. Phase contrast and gradient echo data was acquired using a GE MR750 3T scanner (GEHC, Milwaukee, WI) using a 16 channel wrist coil. Acquisition parameters included: 8 interleave stack of spiral, 512 points per leaf, ±62.5 kHz receiver bandwidth, 10.5 ms TE, 1.9x1.9x4 mm3 voxel size, 128x128x36 matrix, three signal averages, ~15 min scan time. The tracer was captured via quasi-continuous gradient echo scanning acquisition based on the residence time.

Results

The 3D-printing technology (a.k.a., additive manufacturing) of the microfluidic device, vascular network templates and molding jig, enabled fast and convenient prototyping, including user-defined channel geometries, and rapid implementation of the phantom. The 3D-printed, threaded hose-barbs supported fast and secure connections between the device and a pulsatile pump to simulate human heart-beats. 3D printed machine screws ensured the entire phantom was MRI friendly. During experimentation, the phantom supported stable and consistent pulsatile perfusion at physiologically relevant flow and shear stress, that could be tuned to mimic healthy and pathological conditions. The microvascular network was easily visible via phase-contrast and the tracer pulse was captured via gradient-echo.

Discussion

The preliminary data demonstrates the feasibility of simulating biologic perfusion in vitro within a physiologically relevant model, to enable quantitative MRI characterization in healthy and pathologic contexts. Time-resolved tomographic imaging of convective and diffusive tracer transport through tissue is critical to managing stroke, heart attack and cancer, and quantitative interpretation of these data presently uses Kety’s equation, that assumes a global arterial input to all local tissues.2-4 Conversely, this device enables computation of the velocity map from time-resolved image data of tracer transport in vascularized in vitro tissues. This quantitative transport mapping (QTM) approach eliminates a major practical and theoretical difficulty of selecting an arterial input function (AIF) in the traditional Kety’s method for interpreting time-resolved imaging of tracer transport. The resulting quantifiable and reproducible approach provides a promising new method for characterizing blood flow.

Conclusion

The design, fabrication, operation and characterization of a first generation, 3D-printed, physiologically-faithful perfusion phantom has been presented that recapitulates microvasculature structure and function, and enables quantitative analysis of perfusion via MRI. The design provides efficient and high-yield validation of complex and challenging experimental conditions. It sets the stage for the next generation of MRI perfusion characterization via Quantitative Transport Mapping (QTM), that will enable resolution of local transport constitutive equations to time-resolved imaging (4D image data) of tracer flux in in vitro tissues, thus addressing the fundamental global arterial input function (AIF) limit in the current Kety’s method.

Acknowledgements

No acknowledgement found.

References

1. Morgan, J. P. et al. Formation of microvascular networks in vitro. Nat Protoc 8, 1820-1836, doi:nprot.2013.110 [pii]

2. Bammer R. MR and CT Perfusion and Pharmacokinetic Imaging: Clinical Applications and Theoretical Principles: LWW; 2016.

3. Saremi F. Perfusion Imaging in Clinical Practice: A Multimodality Approach to Tissue Perfusion Analysis: LWW; 2015.

4. Barker PB, Golay X, Zaharchuk G, ebrary Inc. Clinical perfusion MRI techniques and applications. Cambridge medicine. Cambridge: Cambridge University Press,; 2013. p xv, 356 p.

Figures

Figure 2 – 3D Printed Microfluidic Perfusion Phantom A) Fully assembled microfluidic perfusion phantom viewed from the side. The phantom consists of a top and bottom piece assembled with 3D printed machine screws. A medical grade silicon O-ring provides a hermetic seal. A type-I collagen matrix is enclosed inside the phantom, containing a microvascular network lined with vascular cells. B.) Phantom viewed from top, showing microvascular network within type I collagen matrix, corresponding to phase-contrast image Fig. 1. C.) Phantom components; top-piece viewed from underside with recess area containing collagen microvascular network (left), top-piece viewed from above showing pump inlet/outlet ports (middle), bottom piece (right).

Figure 3 - 3D-printed Molding jig for Casting PDMS Stamp with Channel Geometry (A) 3D printed mold wafer with channel geometry (right) to cast poly(di-methyl)siloxane, PDMS stamp (left) for formation of collagen channel. Note that cast is a negative image and corresponding PDMS stamp contains a positive, raised feature. B) Fully assembled 3D-printed molding jig (left) and close-up of wafer with channel geometry (right); C.) Top, middle, bottom pieces of Molding jig; D) AutoCAD computer files of 3D print molding jig (3D rendering) corresponding to images in (C). Note that jig is filled with liquid PDMS and cured at 60 degrees Celsius, producing stamp in (A).

Figure 1 – Microvascular Network in 3D-printed Microfluidic Perfusion Phantom A.) Phase contrast image of microvascular network. B.) Fluorescence microscopy image of microvascular network in a type I collagen matrix inside the microfluidic phantom, prior to MRI data acquisition. The vascular network contains human vascular smooth muscle cells and umbilical vein endothelial cells expressing green fluorescent protein. C.) Magnitude image from phase-contrast acquisition. Scale bar 1 mm.

Figure 4 - Time evolution of contrast agent convective and diffusive transport.

Proc. Intl. Soc. Mag. Reson. Med. 27 (2019)

5037