Author: Martin Auerbach, Jeremie Calais, Johannes Czernin, Magnus Dahlbom, Zachary Ells, Adrien Holzgreve, Koichiro Kimura, Karam Masri, Catherine (Caffi) Meyer, David Sennung 👨🔬
Affiliation: Department of Nuclear Medicine, UCLA, UCLA, University of California, Los Angeles 🌍
Purpose: Lutetium-177 PSMA SPECT/CT imaging offers the possibility of qualitative image analysis for tumor response assessment and quantitative image analysis for individualized dosimetry. However, the visual analysis and absorbed dose estimates can be significantly influenced by the SPECT reconstruction parameters. This study aims to evaluate various SPECT reconstruction parameters and their impact on uptake quantitation and clinical image quality.
Methods: This retrospective analysis was performed on Lu177-PSMA-I&T SPECT/CT images acquired in five patients from the LUNAR Trial (NCT05496959). Selection was specific to obtain a representative sample of various tumor sizes, morphologies, and lesion uptake. The images acquired using the Symbia Intevo 16 SPECT at only the 24 hour timepoint post-injection were used. Based on ongoing clinical trials and publications, seven different reconstruction parameters were selected and, four different vendor solutions (MIM Spectra, Siemens xSPECT, Siemens Flash3D, and Hermes Hybrid Recon 4.0) were tested to perform each of the seven reconstructions to assess vendor variability. Four board-certified nuclear medicine physicians blindly assessed the image quality based on noise, contrast, artifacts, resolution, and sharpness using a Likert Scale. The readers also recorded the number of lesions seen on each scan. The SUVmean and SUVmax for each lesion in every scan was determined and evaluated.
Results: A total of 140 scans (7 recons x 4 vendors x 5 patients) were evaluated. There was a total of 12 lesions, the median per patient was 2 [range 1 – 6]. The nosiest images were from 24i32s, highest contrast 8i6s0.9b, greatest resolution 48i1s, and the sharpest image was from 24i32s. The recovery coefficients determined from phantom based measurements varied greatly based on activity concentration, reconstruction parameters, and vendor software used.
Conclusion: Reconstruction parameters and vendor software used has a strong influence on image quality and quantitative accuracy.