Advanced method for dynamic PET imaging using [18F]AIF-NOTA-JR11: For better dynamic imaging of neuroendocrine neoplasms lesions

Introduction: [¹⁸F]AIF-NOTA-JR11 is proved to be a powerful tracer for PET imaging aiming at clinical detection of neuroendocrine neoplasms (NENs). Executing dynamic imaging yields temporally sequential images and can further provides several meaningful parameters based on the subsequent kinetic analysis. However, the quality of dynamic images is limited since the short corresponding scan duration makes the signal-to-noise ratio (SNR) lower than that for a conventional static reconstruction, especially for PET/MR imaging systems where extending scanner axial field-of-view (FOV) is not practical. This is going to challenge the clinical diagnosis and the accurate kinetic analysis based on acquired dynamic images. Therefore, the objective of this study is to propose an advanced method to achieve noise-controlled and accurate 4D (3D spatial + 1D temporal) [¹⁸F]AIF-NOTA-JR11-based dynamic PET imaging, especially for small tumor lesions of NENs.

Methods: The proposed dynamic PET imaging method is based on the application of third-order Hermite interpolation to achieve voxel-level time-activity curves (TACs) correction. TAC smoothing and fidelity is alternately carried out to acquire s optimized balance between image noise controlling and dynamic information preservation. Dynamic PET/MR imaging data acquired by uPMR-790 (United Imaging Healthcare Inc.) using [¹⁸F]AIF-NOTA-JR11 of a patient (44 years old, 62.0 kg, male) with liver NEN were enrolled in this study for the validation of the proposed method. The dynamic scan was performed immediately after an intravenous injection of [¹⁸F]AIF-NOTA-JR11 with the dose of 0.08 mCi/kg. The acquired data were divided into 27 temporal frames (5s×4, 10s×4, 30s×4, 60s×6, 120s×6, 180s×3), and the applied reconstruction method is 3D TOF list-mode ordered-subsets expectation maximization (OSEM) algorithm with 20 subsets and two iterations. Image were reconstructed into a 192×192×115 matrix with the slice thickness of 2.8mm and a transversal FOV of 600mm. Before being processed, the section of the voxel-level TACs corresponding to the early stage were interpolated to have three-fold temporal resolution for better anti-smoothing while the late-stage section is unmodified. Visual inspection and Logan parametric imaging based on plasma input function (PIF) extracted from the dynamic image series were applied to evaluate the performance of noise-controlling of the proposed method and whether undesirable image over-smoothing was introduced, respectively.

Results: The proposed methods is capable to provide JR11 dynamic PET images with higher quality. From visual inspection, the noise level of the processed images is remarkably lower than that of the originally reconstructed images, and clearer imaging of small tumor lesions in the liver with lower contrast with respect with the liver base was achieved. The result image illustrates that the proposed method neither over-smoothed the images nor distorted the dynamic information, the structural information in the original images was well reserved and even optimized.

Conclusions: The proposed method for Dynamic PET imaging using [¹⁸F]AIF-NOTA-JR11 is very effective in terms of noise controlling and the imaging of small NENs lesions without introducing image over-smoothing.

• Download figure
• Open in new tab
• Download powerpoint