Bone material analogues for PET/MRI phantoms

Dharshan Chandramohan, Peng Cao, Misung Han, Hongyu An, John J. Sunderland, Paul E. Kinahan, Richard Laforest, Thomas A. Hope, Peder E. Z. LarsonMedical 2020 · DOI · arXiv

Abstract

Accurate attenuation correction in PET/MRI remains a significant challenge, particularly for bone tissue which exhibits high radiation attenuation but minimal MRI signal. This study investigates the use of plaster of Paris doped with copper sulfate ($CuSO_4$) as a bone material analogue. Results indicate that a mixture containing 2.8% to 4.0% $CuSO_4$ by mass successfully mimics the $T_1$ and $T_2^*$ relaxation times, as well as the 511 keV photon attenuation coefficient of human cortical bone, providing a stable and cost-effective solution for dual-modality phantom development.

Introduction

Simultaneous PET/MRI imaging offers superior soft-tissue contrast and reduced radiation dose compared to PET/CT. However, the lack of a direct relationship between MRI signals and electron density complicates attenuation correction (AC). Bone is particularly problematic; its high density significantly attenuates 511 keV gamma rays, yet its short $T_2^*$ relaxation times (300–500 $\mu s$) result in negligible signal in standard MRI sequences. To validate AC algorithms, physical phantoms that accurately replicate both the radiological and magnetic resonance properties of bone are essential.

Methods

The researchers utilized plaster of Paris (calcium sulfate hemihydrate) as a base material due to its elemental similarity to hydroxyapatite. To tune the MRI properties, three dopants were evaluated: Iohexol (for attenuation), Gadodiamide, and Copper Sulfate ($CuSO_4$).

Samples were prepared in vials with varying concentrations of these agents. Imaging was performed on a 3T MRI system using Ultrashort Echo Time (UTE) sequences to capture the rapidly decaying signals. Attenuation coefficients were measured using CT and $^{68}Ge$ transmission scans. The target benchmarks for cortical bone were defined as $T_1$ (200–250 ms), $T_2^*$ (300–500 $\mu s$), and an attenuation coefficient of 0.10–0.15 $cm^{-1}$.

Results

The study identified $CuSO_4$ as the optimal dopant. While Gadolinium-based agents shortened $T_1$ excessively before reaching target $T_2^*$ values, $CuSO_4$ allowed for more selective tuning of these parameters.

Experimental setup and multi-modal imaging of CuSO4-doped plaster samples.

Plaster doped with 2.8% to 4.0% $CuSO_4$ by mass (measured after a one-week curing period) matched all physiological targets for cortical bone. Undoped plaster was found to naturally possess an attenuation coefficient (~0.14 $cm^{-1}$) within the bone range, requiring only MRI signal adjustment. Stability tests showed that sealed samples maintained these properties for at least 118 days.

Discussion

A critical finding was the impact of dehydration on the material's properties. During the initial week of curing, $T_1$ values increased as water evaporated, while $T_2^*$ remained relatively stable. This suggests that phantoms must be either allowed to reach equilibrium or sealed to prevent further moisture loss. Furthermore, the rapid solidification caused by $CuSO_4$ doping presents challenges for casting complex geometries, such as anthropomorphic skulls, often leading to surface irregularities and trapped microbubbles. Interestingly, these microbubbles were found to further shorten $T_2^*$ through susceptibility effects, inadvertently enhancing the bone-like characteristics of the material.

Conclusion

Plaster doped with copper sulfate is a viable and accessible material for creating PET/MRI bone phantoms. By carefully controlling the dopant concentration and ensuring proper sealing, researchers can produce stable analogues that facilitate the development and validation of advanced attenuation correction techniques in hybrid imaging.

Summary updated

Abstract

Plaster doped with copper sulfate is a promising bone material analogue for a PET/MRI phantom, mimicking the MR properties (T1 and T2*) and 511 keV attenuation coefficient of human cortical bone.
BibTeX
@inproceedings{chandramohan2020bone,
  author = {Dharshan Chandramohan and Peng Cao and Misung Han and Hongyu An and John J. Sunderland and Paul E. Kinahan and Richard Laforest and Thomas A. Hope and Peder E. Z. Larson},
  title = {{Bone material analogues for PET/MRI phantoms}},
  booktitle = {Medical},
  year = {2020},
  doi = {10.1002/mp.14079},
}
Figure