Introduction

Susceptibility artifacts are common in echo planar imaging (EPI) sequences. In vivo imaging of the spinal cord particularly suffers from these artifacts due to susceptibility differences between the various tissue types found in the neck.¹ Standard paramagnetic phantoms that are built for quality assessment in MRI are prepared in cylindrical containers. In this work, we propose a 1:1 scaled agar-agar paramagnetic gel phantom that reflects human head and neck anatomy for quality assessment of MRI protocols.

Methods

3D Design

The 3D design was performed in Autodesk FUSION 360 and Autodesk Meshmixer. Two reference models were used for the body² and the skull and spine³. These models were altered to include skull, cervical spine, thoracic spine, and first two lumbar vertebrae. The body was scaled to match the proportions of an average human of 170 cm height, and the design was cut horizontally under the chest level to have a final height of 44.2 cm, with an 8-mm wall thickness. For the insertion of skull and spine, the design was separated into two interlocking parts 12.2 cm below the top of the head, with 4-mm thick indents created reciprocally on each side to ensure a tight fit (see Fig. 1). The teeth were removed from the model to reduce complexity, and the spine was altered to have a flat surface under the second lumbar vertebrae. Lastly, the skull-spine system was re-scaled to fit into the body (see Fig. 2 and Fig. 3). The body was 3D printed using PETG with 96% infill to ensure water tightness, and the skull-spine system was 3D printed using PLA. The parts were attached to each other using super glue and the exterior interlocking parts were melted together to seal the gaps. Finally, XTC-3D epoxy was used to make the model watertight. Shown in Fig. 2(a), the final size of the phantom was 30.5 x 24.7 x 44.2 cm, with a volume of 14L.

Agar-Agar Gel Preparation

The paramagnetic agar-agar gel was prepared to mimic human white matter T₂, reported as 69 ms at 3T. ^4,5 CuSO₄ was chosen as the paramagnetic agent and NaCl was added to increase conductivity to be similar to that of human tissue.⁶ Agar-agar gel mimics soft human tissues when prepared at appropriate concentrations, where the solidification performance of the gel also depends on concentration. To determine the appropriate concentration, 3 different samples were prepared with 1%,1.5%, and 2% agar-agar in 500 mL containers. The samples were imaged with T₂ -weighted turbo spin echo (TSE) sequence with a series of TE values. The resulting images shown in Fig. 4(a) were used to compute T₂ via fitting to an exponential signal decay curve.⁴ Based on these results, the agar-agar concentration was chosen as 1.5% to achieve slow solidification without saturation. Based on the measured T₂ values from Fig. 4, the required CuSO₄ concentration to match white matter T₂ was adjusted as 4.25mM. For this combination of agar-agar and CuSO₄ concentrations, the T₂ was measured as 61.9 ms, closely matching the target 69 ms.

The gel was prepared in 1L batches with 1.5% agar-agar powder, 4.25 mM of CuSO₄, and 0.2% NaCl.⁷ For each batch, 15 grams of agar-agar and 2 grams of NaCl were mixed with 1L of distilled water, stirred until boiling, and simmered for 2 minutes.⁷ Then, 692 mg of CuSO4 was added while stirring slowly to avoid formation of air bubbles that disrupt the homogeneity of the gel. Mixed every 3 minutes, the gel was allowed to cool down to 55°C to avoid deformities in the phantom, considering the 60°C heat tolerance of PETG. This entire process was repeated in 1L batches until the phantom was filled, with the phantom shaken repetitively to avoid air bubbles. To allow air contact for complete solidification, the phantom was kept uncovered at room temperature for 24 hours before sealing closed.

MRI Acquisition

The phantom was imaged on a Siemens Magnetom 3T scanner with an 8-channel spine array coil. EPI images were acquired in the axial and sagittal planes with 330 mm and 400 mm FOV, respectively. Common parameters were TE= 28 ms, TR=6500 ms, voxel size =2.5x2.5x5 ³ ,16 averages, GRAPPA factor of 4. TSE images were acquired in the axial and sagittal planes with 300 mm and 400 mm FOV, respectively, with TE= 12 ms, TR= 3500 ms, slice thickness = 3 mm.

Results

The final phantom is shown in Fig. 3(b), with the MRI images in Fig. 5. TSE image in the sagittal and axial plane show that the phantom closely matches the human anatomy, while the EPI images show susceptibility artifacts prominent in regions such as the neck where the susceptibility is known to vary rapidly.

Conclusion

We developed a 1:1 scaled agar-agar gel MRI phantom that reflects human anatomy while matching tissue characteristics. The MRI images show that the phantom closely matches human anatomy, and the susceptibility artifacts in EPI images successfully mimic those seen during in vivo imaging of the brain and spine. The phantom will be utilized in the quality assessment of the MRI protocols and image reconstruction techniques in the brain and the spine.

Acknowledgements

This work was supported by the Scientific and Technological Research Council of Turkey (TUBITAK 117E116).