Performance Improvements of the CareMiBrain Brain PET Scanner

Ziel/Aim Our group envisioned in a first step to comprehensively evaluate the performance of the CareMiBrain PET scanner (Oncovision, Valencia, Spain) beyond the NEMA-NU 2-2018 protocol in terms of linearity and homogeneity over the system’s total activity range. In a second step we determined and implemented together with the manufacturer necessary actions to improve the performance.

Methodik/Methods For the linearity evaluation we quantified the image derived activity concentration of a cylindrical phantom in a range between 2.1 MBq to 54.5 MBq. The homogeneity was examined by line profiles. The impact on both entities by a normalization acquired either with a conventional cylindrical or a ring shaped phantom was studied. The same was performed for a new sequential data transfer protocol and an image correction function (ICF) to address inhomogeneity. In order to validate the ICF, measurements with a custom-made grid phantom consisting of vertical and horizontal arranged line sources with distances between 7 mm and 21 mm were performed.

Ergebnisse/Results We found the usage of the ring phantom based normalization in combination with the sequential data transfer protocol yielded a high linearity, with a maximum quantification error of less than 1.50%. Further, the implementation of the ICF led to an improvement of the homogeneity, such as the uniformity of the line profiles, from 16.9±3.2% to 2.6±1.4%. The validation test with the grid phantom was passed within the accuracy limitations of the voxel size.

Schlussfolgerungen/Conclusions Based on the final performance achieved with the system we conducted further tests with the Hoffman brain phantom as well as our first patient studies (n=10). We further aim to evaluate the impact on image quality and accuracy by exploiting the different spatial resolution reconstructions enabled by the monolithic crystal block design. In addition, we are currently working on an enhanced attenuation correction which uses a co-registration with MRI based attenuation maps.

Publication History

Article published online:
14 April 2022