We comprehensively investigated the accuracy and LOD or LOQ of iodine quantification by using fast kilovoltage-switching DECT with DL reconstruction. The relative measurement errors exhibited a decreasing trend beyond a specific nominal iodine concentration. This nominal iodine concentration corresponded to the LOD or LOQ defined in this study (Fig. 2). Additionally, we observed that the measured iodine densities reached a plateau for nominal iodine concentrations for 0.3 mgI/mL in the small phantom and 0.5 mgI/ml in the large phantom, which indicate the limit of the instrument for iodine quantification (Fig. 4). Furthermore, the LOD and LOQ are specific to the instrument used; thus, it is essential to assess these limits along with the accuracy of iodine quantification for specific clinical applications. To the best of our knowledge, this is the first study to perform these evaluations, which are crucial for the clinical implementation of iodine quantification.
When iodine concentrations exceed the LOD or LOQ, both the relative measurement errors and RSD presented a decreasing trend as the nominal iodine concentration increased, in both small and large phantoms, irrespective of CTDIvol. This finding suggests that higher target iodine concentrations can be measured with greater repeatability, which leads to enhanced accuracy and repeatability. No such trend was observed for iodine concentrations below LOD or LOQ. In particular, measurements for nominal iodine concentrations below the LOD were indistinguishable from measurements of 0.0 mgI/mL (no iodine) and may lead to misdiagnosis by iodine density in clinical practice, despite their high accuracy. For the phantoms that contained seven iodine inserts used in this study, the LOD and LOQ that were obtained from the larger phantoms were approximately twice as large as those obtained from the smaller phantoms (Table 1), and the LOD and LOQ decreased as the CTDI increased for both small and large phantoms. Because the LOD and LOQ were phantom size and CTDIvol dependent, it was necessary to adjust the CTDIvol according to the target organ.
Compared with prior research [13], this study obtained higher LOD and LOQ values, which may be due to the 20 samples minimum required to obtain the LOD and LOQ values [16]. The discrepancy in the sample size between our study, which derived LOD and LOQ from 30 measurements, by ten scans across all three consecutive cross-sections, and the prior study, which used five measurements, is considered as one of the factors for the variation in the results between the two investigations. The accuracy and RSD improved at higher nominal iodine concentrations, and the number of measurements had a significant impact on the measured iodine densities at low nominal iodine concentrations. Therefore, this study, with its large sample size, might have evaluated the LOD and LOQ more precisely when low iodine concentrations were investigated. Furthermore, the prior research computed LOD and LOQ based on a higher CTDIvol than that employed in clinical settings to disregard the noise effect. Thus, our study offers better insights into the accuracy and LOD/LOQ at the CTDIvol used in clinical settings.
This study has a few limitations. First, all image reconstructions were performed using the standard DECT reconstruction parameters. However, recent research has revealed that the accuracy of iodine quantification provided by body parameters surpasses that obtained using other parameters [15]. Second, the phantoms used in this study were non-anthropomorphic and represent a simplified simulation of patient size that did not incorporate anatomical structures that influence the accuracy of iodine quantification, such as the spine or other organs. Therefore, variations in our findings may arise when applied to actual patients. Finally, our study exclusively employed a fast kilovoltage-switching DECT scanner with DL reconstruction from a single vendor, and whether the results can be replicated using other DECT scanners remains undetermined. Therefore, future investigations should evaluate other clinically available DECT implementation methods, such as dual-source CT, dual-layer CT, and twin-beam CT, owing to instrument-specific LOD and LOQ values.