DCE MRI provides information about Pharmacokinetics (PK) in tissues which aids in detection, staging and looking at the behaviour of vascularized tumours. Current work focuses on the design and development of in vivo DCE phantom for providing a reference standard for quantitative validation and the ability to generate Signal Intensity (SI) curves that mimic Arterial Input Functions (AIF) of different anatomies.

The DCE MRI phantom setup is shown in the figure 1. The phantom has three acrylic tubes (diameter: 2mm, 2mm and 3mm) for water flow which mimics the arteries. Small pores of 1mm diameter was made at the centre of all three tubes to mimic leakiness of Contrast Agent (CA) to the tissue mimicking material. The three tubes were connected to each other making a single inlet and outlet as shown in figure 1(a-b). Further the inlet was divided into two tubes by using the hose, where one of tubes is connected to the main water source providing constant water flow and the other tube was used to inject the CA using a syringe. Copper sulphate (CuSO₄ 0.25g/100ml) was used as the CA. The water outlet was connected to a bucket outside the scanner.

Tissue mimicking was performed using Poly Vinyl Alcohol (PVA)¹.10% of PVA solution was prepared by boiling PVA salt in water at 100⁰ C. The mixture was heated until the solution was clear to obtain the PVA cryogel. The PVA cryogel was cooled gradually to room temperate with the lid closed so that the water content in cryogel is intact. The PVA cryogel was poured into a container and placed at -18⁰ C for 3 freeze-thaw cycles. Multiple pores were made to make the material porous in order to obtain the wash-in and wash-out curves. The set-up box was made from acrylic plastic and was machined in house. The central part of the three tubes were then covered with the PVA material thus prepared which can be seen in figure 1(b). The total bill of material for all the components used to build the phantom was less than USD $75 and was assembled with readily available material and tooling. Phantom experiments were conducted on 1.5T Siemens MR scanner to obtain the SI curves. Images were acquired using the 3D gradient recalled acquisition with parameters of TR/TE=8.11/4.76 ms, number of average=1, 20 slices with 30 time points with temporal resolution of 12 s. The total scan time was about 6 mins with the CA being injected after 1 min. SI curves were plotted to depict the wash-in and wash-out characteristics of the phantom for the three different perfusion components (seen in figure 1) with the one in the middle being able to let the flow through at a constant rate. This served as the control for the other two components. Tissue mimicking PVA was needle pricked to enable porosity around the ROI₁ and ROI₃ region to ensure that CA leaks out rapidly from that particular region.

Figure 2(b) depicts the SI plots for the three different regions of tissue marked with regions of interest in figure 2(a) with corresponding colors. It can be observed that SI curves for the two ROIs were able to produce changes in SI with respect to the control over the time for which data was being acquired. The curves are similar to arterial input functions with different rates similar to that of real in vivo data where there is a rapid wash in and wash out depicted in Figure 2(c) for population averaged AIF². It can be observed that the tube with 3mm diameter (ROI ₃) resulted in increased flow rate as is expected with the signal change rate of of R₁=18.71 (change in signal intensity /s) for ROI₃, R₂=5.11 for ROI₁ and R₃=1.26 for ROI₂.A perfusion phantom that mimics the AIF curve has been developed at low cost. The increased flow rate in tube 3 can be attributed due to the increase in the diameter of the tube. The SI profile obtained from ROI₂ is different to the other two ROIs due to the absence of hand-made/needle pricked pores. In conclusion, a SI curve similar to that of AIF was produced. Future studies of the phantom would be to mimic the different tumour tissue curves, where there is a sudden wash in and slow wash out. This would be then subject to application of relevant PK modelling based on the understanding of the nature of perfusion of the tumour type.