Design of a realistic PET-CT-MRI phantom

Abstract

The validation of the PET image quality of new PET-MRI systems should be done against the image quality of currently available PET-CT systems. This includes the validation of new attenuation correction methods. Such validation studies should preferentially be done using a phantom. There are currently no phantoms that have a realistic appearance on PET, CT and MRI. In this work we present the design and evaluation of such a phantom. The four most important tissue types for attenuation correction are air, lung, soft tissue and bone. An attenuation correction phantom should therefore contain these four tissue types. As it is difficult to mimic bone and lung on all three modalities using a synthetic material, we propose the use of biological material obtained from cadavers. For the lung section a lobe of a pig lung was used. It was excised and inflated using a ventilator. For the bone section the middle section of a bovine femur was used. Both parts were fixed inside a PMMA cylinder with radius 10 cm. The phantom was filled with <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">18</sup> F-FDG and two hot spheres and one cold sphere were added. First a PET scan was acquired on a PET-CT system. Subsequently, a transmission measurement and a CT acquisition were done on the same system. Afterwards, the phantom was moved to the MRI facility and a UTE-MRI was acquired. Average CT values and MRI R <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sub> values in bone and lung were calculated to evaluate the realistic appearance of the phantom on both modalities. The PET data was reconstructed with CT-based, transmission-based and MRI-based attenuation correction. The activity in the hot and cold spheres in the images reconstructed using transmission-based and MRI-based attenuation correction was compared to the reconstructed activity using CT-based attenuation correction. The average CT values in lung and bone were -630 HU and 1300 HU respectively. The average R <sub xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">2</sub> values were 0.7 ms <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">-1</sup> and 1.05 ms <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">-1</sup> respectively. These values are comparable to the values observed in clinical data sets. Transmission-based and MRI-based attenuation correction yielded an average difference with CT- based attenuation correction in the hot spots of -22 % and -8 %. In the cold spot the average differences were +3 % and -8 %. The construction of a PET-CT-MRI phantom was described. The phantom has a realistic appearance on all three modalities. It was used to evaluate two attenuation correction methods for PET-MRI scanners.