Quantification of the F18-FPN brain PET images by Reference Tissue Model

Objectives: F-18-Florapronal (FPN, Alzavue, FutureChem Pharma, Korea) is a developed radio-pharmaceutical for amyloid brain PET imaging. The aim of this study is to calculate early and parametric images (R1, BPnd) from dynamic F-18-FPN brain PET images and use them for AD-HC classification. Materials and Methods: Seventeen AD patients (age: 69.0±9.3 yrs) and 15 age-matched normal subjects (age: 69.9±8.0 yrs) underwent F-18-FPN brain PET scans. Dynamic brain PET images (32 frames) were acquired 0-60 min after intravenous injection of F-18-FPN (370 MBq). Early-phase F-18-FPN (eFPN) PET images were merged by frame over 1.5-4 minutes. Region-based statistical comparisons were calculated using the average SUVr values ​​in 83 brain regions using Hammer’s brain atlas of PMOD 3.6. R1 and BPnd parametric images were calculated using the SRTM, SRTM2, and MRTM2 models by selecting the receptor-rich region (precuneus, posterior cingulate, inferior parietal, lateral temporal cortices) and receptor-less region (cerebellum gray matter). The values of parametric ​​images for each area listed in ​​Hammer's atlas were extracted and a statistical comparison was made by employing the MWU-test. The optimal sensitivity and specificity were found by generating ROC, and the cut-off value was found after calculating the Youden index.

Results: In the R1 parametric images, 12 of the 83 regions listed in Hammer's atlas showed a decrease in blood flow in the AD group. The regions of decreased blood flow in the AD group were amygdala, lateral ventricle, subcallosal area, third ventricle, parahippocampal and ambient gyri, caudate nucleus, thalamus, and nucleus accumbens. The ROC for AD-HC classification using the value of the lateral ventricle region in the R1 parametric image had an AUC of 0.819, sensitivity 82.35, specificity 92.86, and cut-off value 0.35. The ROC for AD-HC classification using the SUVr value of the Amygadala region in the eFPN image had an AUC of 0.825, sensitivity 80.0, specificity 87.5, and cut-off value 0.87. BPnd parametric images of AD and HC groups were statistically different in the frontal cortex, lateral temporal cortex, parietal cortex, posterior cingulate cortex, cerebella cortex, and gray matter areas with p<0.001. The performance of AD-HC classification using R1 parametric image was similar to that using eFPN image. In ROC for AD-HC classification using the BPnd value of the parietal lobe gray matter region, the AUC was 0.996, sensitivity 93.75, specificity 100.0, and cut-off value 0.11. Conclusion: The classification of AD-HC was clearer when comparing values extracted from each sub-unit region listed in Hammer's atlas in the R1 parametric image than when comparing the values in a large unit region in the BPnd parametric image. It is suggested to use both R1 parametric image and eFPN image for AD-HC classification for making blood flow comparison.