Design, Construction and Evaluation of an Anthropomorphic Head Phantom for Assessment of Image Distortion in Stereotactic Radiosurgery Planning Systems

Introduction:
In recent years, the use of magnetic resonance (MR) images in radiation treatment planning has drawn considerable attention. However, although the extent of a tumor can be determined in great detail on MR images, the geometric accuracy of these images is limited by distortions stemming from the inhomogeneity of the static background magnetic field, the nonlinearity of the applied gradient magnetic fields, the magnetic susceptibility of the imaged tissues, and chemical shift artifacts. Our goal is to design and construct a new Anthropomorphic Head Phantom for Assessment of Image Distortion in Treatment Planning Systems.
Materials and Methods:
In this study, CT scan images of head were transferred to the Mimic software (Mimics® Innovation Suite). Using this software, the skull texture was removed and a hollow layer formed between the bone tissue in which the bone tissue would be equivalent to the material. Then fabricated it with a 3D printer and used K2HPO4 (as bone). we designed a new 3D grid containing 13,824 reference features (control points) with AutoCAD software, fabricated it with a 3D printer, and filled it with gels that include nickel-doped agarose, sucrose, urea, and sodium chloride (as soft tissue) then placed this grid Inside the head phantom. The phantom is supplied with specially designed pads that allow fixation with any stereotactic frame or mounting for end-to-end testing. This phantom was tested on the Siemens 3 Tesla Prisma MRI model using a 64-channel head coil. We used imaging from a six-slice CT scan (Siemens) as a reference and matched the reference features in the MRI images with the CT-scan images. To achieve this, we used a three-dimensional reference feature model. Reproducibility on the phantom was investigated with three different imaging sessions per day for three different days.
Results:
The CT numbers and relative electron density of the gels closely resembled those of real tissue and T1 and T2 of the gels closely resembled those of real tissue. The geometric distortion in the 3D results was found to be due to field non-uniformity and nonlinearity of the gradients and its reproducibility. The mean Euclidean distance error for MRI volume was less than 0.9 mm. The maximum Euclidean error was 1.3 mm. Distortion in the whole volume is pronounced exclusively at the edges of the magnetic field.
Conclusion: As with other investigations carried out in our work, the amount of distortion in the middle of the field was less than at its sides. This phantom can be used to check Image Distortion in Treatment Planning Systems.