Abstract
1001
Objectives: Most SPECT acquisition protocols are defined with the assumption of finding and characterizing abnormal patterns in the diagnostic stage of imaging. That mandates a certain count level to keep statistical uncertainty at bay to perform that diagnostic task. However, in a theranostic setting using Lu-177-PSMA-ligand for treatment, and F-18-PSMA-ligand for the diagnostic imaging task, the SPECT acquisition can be sped up. Clinically, only a small selection of representative target lesions is relevant. It is mandatory to have robust identification of change against baseline and thus a standard’s referenced calibration of the imaging system. The goal is to find a 3 FOV SPECT protocol with adequate statistical uncertainty to perform a clinically relevant comparative assessment of change in quantitative tumor uptake at 2 time points to monitor treatment response in less than 30 min.
Methods: We first selected 1 patient data set acquired with a calibrated Symbia Intevo16 (“classical”; 45 minutes dwell time) at 2 different time points with known tumor locations from F18-PSMA image. Clinicians defined 4 volume-of-interest (VOI) of 3, 4, 6 and 11 ml to be used as proxy during follow up (Figure 1). We then binomially subsampled the projection data at different reduction levels with research software and subsequently reconstructed with xSPECT Quant [1, 2]. xSPECT Quant yields an estimate of the quantitative tracer uptake from a calibrated SPECT system using a 3%-NIST traceable Se-75 calibration source. We reconstructed the variance images from 10 noise realizations using bootstrapping and analyzed the clinically relevant quantitative uptake and uncertainty of VOIs in the thorax, abdomen and pelvis for 2 time points using responsiveness analysis. An unpaired t-test is used to assess means difference and confidence interval of differences. Visual inspection of the resulting images is performed to assure that the quantitative images also maintain visually extractable clinical information. The reconstruction parameters (48it, 1ss, 10mm) were kept identical to the “classical” baseline scan to simplify clinical workflow, accepting the tradeoff of about 10% lower mean uptake as compared to the “classical”. Results: We assessed a variety of time reduction cases and show here the result for the border case of a dwell time of 10min for all 3 bed positions. We computed the difference between July (T2) and May (T1) for each noise realization pair and count reduction case leading to 100 pairs per reduction case. We see that between both time points uptake decreased in VOI2 and VOI3 but increased in VOI1 and VOI4. We show analysis results both for VOI’s stacked (left) and individually (right) in Figure 2 only for the 10min dwell time case. The t-test cannot reject the null hypothesis and at p>0.9. Images were compared visually as well as the VOI mean uptake with ±1SD. At 5s/view; 60 views/detector and continuous acquisition the means in the selected VOI’s were not statically significant to each other. Conclusion: For the quantitative comparative task assessing treatment response of 7GBq injections of Lu-177-PSMA between 2 time points we developed a wholebody (eye-to-thigh) acquisition protocol with a dwell time of 3.33 min per bed for a calibrated Symbia Intevo16. The clinic has decided to increase the dwell time to 15min resulting in a 20min protocol (5s/view; 60views/detector; continuous NCO; 180 deg. configuration;180-degree rotation) to image over 3 FOVs. This protocol has been in practice for about 6 months on about 50 patients (Figure 3). It was found suitable for clinical work and to assess tumor burden change over multiple time points. Reference:1. Vija, A.H., von Gall C., Ghosh P., Accurate, Reproducible and Standardized Quantification, in Molecular Imaging White Paper, 2015, .2. Vija, A.H., Characteristics of the xSPECT reconstruction method. White Paper, 2017. Siemens Medical Solutions USA, Inc., Molecular Imaging
In this issue
Jump to section
Related Articles
Cited By...
- No citing articles found.