Emerging research suggests that MRI-guided reconstruction may facilitate improved evaluation of amyloid brain positron emission tomography (PET) scans.
For the retrospective multireader study, recently published in Radiology, researchers comparedordered-subset expectation-maximization (OSEM) reconstruction of PET imaging versus MRI-guided PET reconstructions for 120 patients (median age of 74.8) who had brain PET/MRI scans with (18F)-florbetapir.
The study authors found that MRI-guided PET reconstruction enabled a 12 percent higher inter-reader agreement in contrast to OSEM reconstruction (91 percent vs. 79 percent).
MRI-guided PET reconstruction offered significantly higher accuracy (97.9 percent vs. 94.1 percent), sensitivity (97.1 vs. 95.2 percent) and specificity (99.2 percent vs. 91.6 percent), according to the researchers. The researchers also noted a 7.3 percent reduction in the false positive rate (0.8 percent vs. 8.1 percent) with MRI-guided PET reconstruction.
“MRI-guided reconstruction improved diagnostic performance, driven by increased specificity and fewer false positives. This is clinically important as it may reduce unnecessary treatment and follow-up,” noted lead study author Yu Cai, MD, PhD, who is affiliated with the Nurturing Center of Jiangsu Province for State Laboratory of AI Imaging and Interventional Radiology within the Department of Radiology at Zhongda Hospital in Nanjing, China, and colleagues.
Three Key Takeaways
• Higher diagnostic accuracy and reader agreement. MRI-guided PET reconstruction improved inter-reader agreement by 12 percent (91 percent vs. 79 percent) and boosted overall accuracy (97.9 percent vs. 94.1 percent), sensitivity (97.1 percent vs. 95.2 percent), and especially specificity (99.2 percent vs. 91.6 percent) compared to standard OSEM PET reconstruction.
• Fewer false positives, less unnecessary follow-up. The 7.3-percentage-point drop in false positive rate (0.8 percent vs. 8.1 percent) suggests MRI-guided reconstruction could reduce unnecessary treatment or follow-up testing driven by misread amyloid scans.
• Practical tradeoff with the time cost. Despite the diagnostic gains, MRI-guided reconstruction takes about two hours per scan, a workflow limitation that may hinder routine clinical adoption.
The researchers suggested that the differentiation of gray and white matter may be a key benefit with the use of MRI-guided reconstruction in evaluating amyloid PET scans.
“MRI-guided reconstruction incorporates MRI-derived anatomic priors that enhance gray-white matter contrast, central to amyloid PET visual interpretation. However, anatomic priors can introduce non-shared gradient transfer or suppress PET specific features when structural constraints are strongly enforced. These confounding effects are less of a concern in amyloid imaging as the primary gray-white matter gradient is shared between the modalities except in clearly positive scans with high gray matter amyloid burden,” posited Cai and colleagues.
(Editor’s note: For related content, see “FDA Clears NeuroQuant PET for Assessment of Neurological Conditions,” “A Closer Look at the Evolution and Emerging Insights with Amyloid and Tau PET for Dementia and Alzheimer’s Disease” and “Can DSC MRI-Derived Oxygen Ejection Fraction Predict Diffusion Reversal in Patients Treated for Acute Ischemic Stroke?”)
Beyond the inherent limitations of a single-center retrospective study, the study authors conceded potential bias due to a lack of true blinding to the image reconstruction method employed and use of a reference standard based on unanimous OSEM evaluations from the reviewing readers. The researchers also noted the lack of efficiency with MRI-guided PET reconstruction taking two hours per scan.
