Development of a technique based on multivariate statistical method to enhance scanning time and lesion detectability in dynamic PET

Abstract

Dynamic PET imaging plays a key role in improving cancer diagnosis, assessing therapy response, and characterizing tumor lesions. However, dynamic PET suffers from several limitations such as noisy images and longer data acquisition time. This study aimed to enhance lesion detectability in 18F-FDG PET images and shorten the total duration of the 18F-FDG study using principal component analysis (PCA), while preserving the detectability of lesions. For this purpose, activity maps were generated using the 4D-XCAT anthropomorphic phantom combined with time activity curves calculated using a standard 18F-FDG 3-compartment model. A 9 mm spherical liver lesion was inserted into the 4D-XACT phantom. The widely used STIR image reconstruction software was used to generate dynamic 18F-FDG PET images. The PCA method was applied to the simulated dynamic 18F-FDG-PET images generated at different scanning times (0-11, 0-13, 0-15, 0-20, 0-25, 0-35, 0-40, 0-45, 0-50, and 0-55 min) and at T=35-55 min. The estimated Principal Component Images (PCIs) were visually assessed and compared to the SUMmed images generated at T=35-55 min. The Tumor-to-Background Ratio (TBR) and Activity Profiles Line (APLs) were considered in the quantitative assessment. For lesion detectability, the comparison between the 3PCIs (PCI1, PCI2 and PCI3) and SUM images showed that PCI2 could provide superior lesion detectability compared to the SUM images at T-35-55. Concerning the time acquisition reduction, the visual assessment of the 3 estimated PCIs and SUM images showed that PCI3 has better detectability of the lesion regardless of the scanning time. PCI2 showed good lesion detectability at all timing expect at 11 and 13 min, whereas PCI1 showed lower lesion detectability at all times. Quantitatively, the TBR estimated on PCI3 was similar to that on the SUM images (TBR on PCI3 was 2.49±0.72 and 3.08±0.49 on the SUM images) but higher than that on PCI1 and PCI2 from timing 11 to 35 min. The TBR on PCI1 decreased with the increase in scanning time. On PCI2, the TBR increased with scanning time (maximum of 3.87±1.11 at 55 min). The findings of this study suggest the use of the PCI3 at 20 and 25 min to resucethe scanning time in dynamic 18F-FDG PET. This study demonstrated that PCA allows enhancement lesion detectability and reduction of scanning time in dynamic 18F-FDG-PET imaging, hence increasing patient comfort and minimizing motion effects