In this study, the properties of SwiftScan in quantitative bone SPECT were investigated using a phantom. The CV and CNR of SwiftScan tended to be superior to those of SSM in short acquisition time. However, in long acquisition times, CV of SwiftScan and SSM were comparable, whereas the CNR of SSM was higher than that of SwiftScan. SwiftScan and SSM showed almost the same RC, regardless of the acquisition time in terms of quantitative performance. These results suggest that SwiftScan provides the same or better image quality and quantitativeness as SSM in short acquisition time quantitative bone SPECT examination.
The CV of SwiftScan was lower than that of SSM for short-time acquisition, but they were almost equal for long-time acquisition. The reason for this can be explained by the fact that SwiftScan adds the count of detector moving time (approximately 4 s/view, total 120 s) to the SSM. When the SSM is at 5 min acquisition (6 s/view for SSM), SwiftScan increases the count by 1.67 times (6 s/view + 4 s/view), whereas when the SSM is at 32 min acquisition (60 s/view), SwiftScan only increases the count by about 1.07 times (60 s/view + 4 s/view). Cao et al. [22] compared the image quality of simulated continuous SSM (CSSM) with that of SSM through simulation experiments, and they showed that CSSM is almost the same as SSM for long-time acquisition, but CSSM substantially improved for short-time acquisition.
The CNR of SwiftScan was higher than that of SSM in short-time acquisition due to improved contrast in the tumor area. However, the CNR of SSM was equivocal or higher than that of SwiftScan in long-time acquisition. The reason for this is the same as that for CV: in the short-time acquisition, the effect of the added counts during detector movement in SwiftScan is significant. The acquisition parameters, such as the number of projections and acquisition time, are less relevant in regions with high-count statistics (e.g., in the tumor area) compared with regions with low-count statistics (e.g., in the background) [13]. Therefore, SwiftScan is more advantageous than SSM in short-time acquisition because the image quality can be improved by adding the counts of detector moving time. However, for long acquisition time, there might be no advantage in using SwiftScan instead of SSM.
In the FWHM evaluation, SSM and SwiftScan deviated from the actual size (10 mm) at 5 min acquisition (6 s/view). This deviation may be owing to insufficient convergence of the reconstruction owing to high noise in the short acquisition time. Furthermore, SwiftScan may have increased the FWHM due to the addition of counts during detector movement. Such an increase in FWHM (degradation of resolution) may affect the region of interest settings and the reliability of quantitative value.
In quantitative value comparison, SwiftScan and SSM were similar. However, in both cases, the RC was higher for smaller tumor diameters in short-time acquisition. The quantitative analysis software accompanying the SPECT/CT system used in this study was designed to calculate quantitative values without noise reduction filters. Therefore, the noise reduction filter was not used for all SPECT images. Thus, the quantitative values may have been estimated high due to the extremely high noise in the short-time acquisition and the significant effect of noise in the ROI due to the small tumor diameter. The short-time acquisition images without noise reduction are insufficient for reading by diagnosticians due to the high noise. Kupitz et al. [13] recommended using two different reconstruction protocols with task-specific optimized setups (quantitative and non-quantitative). Based on the results of CV and CNR, the number of iterations was set to 5. It is said that increasing the number of iterations converges the quantitative values to a known value [23]. In this study, the quantitative value of the reference and other tumor parts did not change with the number of iterations (data are not shown). However, note that an increase in the number of iterations increases noise.
In quantitative bone SPECT, attempts to reduce acquisition time are being investigated. Ichikawa et al. [24] showed that the xSPECT bone reconstruction algorithm could maintain high image quality and quantitativeness even with a 3-min acquisition time. Although the reconstruction algorithm currently depends on the vendor, there is a possibility of further shortening of acquisition time and/or improving image quality and quantitativeness by applying a reconstruction algorithm, such as xSPECT bone to SwiftScan. Moreover, SwiftScan may reduce the total examination time if the image quality and quantitativeness are sufficient. Shortening the total examination time and obtaining quantitative and image quality comparable to that of SSM would reduce the possibility of patient motion and mental and physical distress.
This study has several limitations. First, the effect of the change in the acquisition time and radioactive concentration level has not been investigated in detail because it is a phantom experiment. Therefore, it is necessary to investigate the detailed acquisition time variation by simulation. Second, the lack of comparison with the CM acquisition. Since continuous acquisition is used in many SPECT/CT devices, a comparison is necessary. Lastly, there is a lack of clinical study. It is necessary to examine whether the results of phantom experiments are equivalent to those in clinical practice.