PET image spatial normalization method may reduce interpretation variability, boost Alzheimer’s research
Researchers from Sweden have developed a fully automated PET spatial normalization method for amyloid-β (Aβ) imaging that may reduce interpretation variability among readers and monitor the effectiveness of patients treated with anti-Aβ drugs—a promising step for Alzheimer’s research.
Lead author Johan Lilja, PhD, with Uppsala University in Sweden and colleagues studied 117 total patients in research published online June 14 in the Journal of Nuclear Medicine.
Currently, only three amyloid-β specific fluorine-18 radioisotopes are approved by the FDA and European Medicines Agency—including 18F-flutemetamol. These are only approved for visual assessment whereby images are classified by a reader, but there are data inferring an automated approach may be superior, researchers noted.
Furthermore, the process for quantification is long and arduous, requiring PET images to be spatially normalized, access to a patient’s T1-weighted MR image, determination of a standardized uptake value ratio (SUVR) and a manual delineation of volumes of interest (VOI), Lilja et al. wrote.
“This method, however, is time consuming and may be subject to inter-reader variability; further, structural imaging with can prove challenging to obtain in clinical settings,” authors wrote. “As such, PET-only approaches have been developed, where, following normalization to a reference space, division of the Aβ image into VOIs and subsequent calculation of SUVR is achieved using a computer generated, predefined regional atlas.”
The participants were divided into two groups—a template-creation cohort of 70 participants from a 18F-flutemetamol phase II study and a registration validation group of 47 patients from an ongoing study at a Swedish hospital investigating the PET tracer’s effectiveness in patients with an unclear diagnosis.
Lilja and colleagues used an MR-driven registration method (SPM12) to spatially normalize 18F-flutemetamol PET and corresponding MR images that were ultimately used to create a synthetic template which authors incorporated into their image registration method. This became their PET only registration method.
The new technique was evaluated using the lengthy PET and MR combination method authors previously described as “time consuming.”
Results showed all scans were spatially normalized using the newly proposed method and required no manual adjustments. 18F-flutemetamol quantification revealed “strong agreement” between the SUVR values for the PET and MR method, according to Lilja et al.
“The principal component template registration method allows for robust and accurate registration of [18F]flutemetamol images to a standardized template space, without the need for an MR image,” authors wrote. “Automated quantification using the proposed method might increase reader certainty and further the clinical adoption of Aβ imaging, including within the context of clinical trials.
Matt joined Chicago’s TriMed team in 2018 covering all areas of health imaging after two years reporting on the hospital field. He holds a bachelor’s in English from UIC, and enjoys a good cup of coffee and an interesting documentary.