Introducing a CT tin filter into SPECT CT acquisitions: an evaluation of the effect on Hounsfield units, SPECT quantification accuracy and patient dose.

Introduction: A tin (Sn) filter has been introduced to the Siemens Symbia Pro.specta SPECT-CT system. Used previously on standalone CT systems and recently introduced to PET-CT systems, Sn filters have been shown to significantly reduce patient dose with minimal compromise on image quality. In this paper, phantom work has been undertaken to establish the impact of a Sn filter on SPECT-CT acquisitions. Firstly, the effect on Hounsfield units (HU) was evaluated. The impact on quantitative SPECT reconstruction accuracy was assessed for both Siemens Broad Quantification™ and Hermes SUV SPECT® methods for a range of radioisotopes: Tc-99m, I-123 and Lu-177. The level of patient dose saving that could be achieved when using the Sn filter was then established by analysing CT image noise versus patient dose.

Methods: A Multi Energy CT phantom (Gammex model 1472) was used for this project. Ten tissue and contrast equivalent inserts were placed in the phantom to allow evaluation of HU. Three 20ml syringes were inserted in the phantom to look at quantitative reconstruction accuracy for Tc-99m, I-123 and Lu-177 in turn. The syringes were filled with approximately 5MBq, 10MBq and 80MBq.

For each radioisotope assessed, the phantom underwent 1 SPECT scan and 60 CT scans, 30 acquired with the Sn filter turned on and 30 with the Sn filter turned off. A series of different kV and mAs settings were used for the CT in order to evaluate a range of possible clinical protocols. kV was varied between 100-140kV and mAs between 20-200mAs.

The impact of the Sn filter on HU was calculated by selecting 7 different regions of the Multi Energy CT phantom: air and various tissue and contrast objects. The HU in each region was measured on the Sn filter CT image and compared to the non-Sn filter CT image for each kV and mAs combination used.

Accuracy of quantification was assessed by measuring the activity recovered in each syringe for each radioisotope tested from quantitative SPECT reconstructions. Both Siemens Broad Quantification™ and Hermes SUV SPECT® quantification reconstructions were produced and evaluated. Percentage difference was calculated between activity recovered from the reconstructions carried out with the Sn filter and the equivalent reconstructions carried out without the Sn filter.

To look at CT patient dose versus image quality with and without the Sn filter, the CTDIvol for each study was plotted against noise measured in the background region of a phantom central slice for each CT scan.

Results: HU are lower when using the Sn filter due to beam hardening, results found are in agreement with previous literature on the use of a Sn filter in standalone CT applications. One small, non linear effect was noted happening at CT numbers between 50 and 150, however this is not a concern for the purpose of AC CT.

For quantitative SPECT reconstructions using Siemens Broad Quantification™, the difference in activity recovered was less than 1% for all kV and mAs combinations in all but one case where a difference of -4% was found. Using Hermes SUV SPECT® quantitative reconstructions, the difference in activity recovered was less than 1.5% for all but 3 results, where the largest difference measured was 5%.

When using the Sn filter, on average, a CT dose saving of 60% was achieved for the same level of image noise. If additional image noise can be tolerated, then further dose reduction is possible. For the standard clinical protocol used for patients of 120kV and 60mAs, a dose saving of 75% was achieved.

Conclusions: The Sn filter does not have a significant impact on HU. Minimal differences were found between quantitative SPECT reconstructions performed for scans acquired with and without a Sn filter for both Siemens Broad Quantification™ and Hermes SUV-SPECT® methods. Significant patient CT dose savings can be achieved using the Sn filter with little compromise on image quality.