UCSF

Purpose

To demonstrate the use of anatomic MRI-visible three-dimensional (3D)-printed phantoms and to assess process accuracy and material MR signal properties.

Methods

A cervical spine model was generated from computed tomography (CT) data and 3D-printed using an MR signal-generating material. Printed phantom accuracy and signal characteristics were assessed using 120 kVp CT and 3 Tesla (T) MR imaging. The MR relaxation rates and diffusion coefficient of the fabricated phantom were measured and ¹ H spectra were acquired to provide insight into the nature of the proton signal. Finally, T₂ -weighted imaging was performed during cryoablation of the model.

Results

The printed model produced a CT signal of 102 ± 8 Hounsfield unit, and an MR signal roughly 1/3^rd that of saline in short echo time/short repetition time GRE MRI (456 ± 36 versus 1526 ± 121 arbitrary signal units). Compared with the model designed from the in vivo CT scan, the printed model differed by 0.13 ± 0.11 mm in CT, and 0.62 ± 0.28 mm in MR. The printed material had T₂ ∼32 ms, T2*∼7 ms, T₁ ∼193 ms, and a very small diffusion coefficient less than olive oil. MRI monitoring of the cryoablation demonstrated iceball formation similar to an in vivo procedure.

Conclusion

Current 3D printing technology can be used to print anatomically accurate phantoms that can be imaged by both CT and MRI. Such models can be used to simulate MRI-guided interventions such as cryosurgeries. Future development of the proposed technique can potentially lead to printed models that depict different tissues and anatomical structures with different MR signal characteristics. Magn Reson Med 77:613-622, 2017. © 2016 International Society for Magnetic Resonance in Medicine.