Impact of patient habitus and acquisition protocol on iodine quantification in dual source photon-counting CT

Introduction

Intravenous administration of iodinated contrast during CT acquisitions can improve diagnostic utility for many clinical indications by improving contrast between structures with similar attenuation on non-contrast (native) scans. In addition to visual assessment of contrast enhancement, iodine measurements provide the ability to assess contrast uptake of various tissues, organs, and lesions quantitatively. However, CT systems are only capable of assessing the amount of iodine at best in a semi-quantitative way as highly attenuating materials, such as calcifications or bone, can have HU values in the same range as iodine. With beam hardening artifacts present in conventional CT, these materials can make it difficult for contrast-enhancing features to be differentiated from non-enhancing features.

Spectral CT can overcome challenges of conventional CT, such as beam hardening or non-quantitative measurements, in order to accurately determine iodine levels using iodine density maps. The latest addition to the family of spectral CT instruments is photon-counting computed tomography (PCCT)¹. Many years of preclinical research have demonstrated superiority of photon-counting detectors (PCDs) over energy-integrating detectors (EIDs) in image quality (high contrast-to-noise ratio, high spatial resolution, low electronic noise, and reduced artifacts), quantitative imaging, and dose efficiency^2–8. Since PCCT has made the transition into the clinical routine⁹, iodine quantification studies on first-generation clinical systems have further confirmed its superiority over EID-CT in applications like abdominal^10–12, and cardiac imaging^13–15.

Accurate iodine quantification is needed to ensure reliable clinical results. Consequently, the purpose of this study was to characterize iodine quantification performance of a first-generation clinical dual-source PCCT system (NAEOTOM Alpha, Siemens Healthineers). A phantom containing iodine inserts of various concentrations was utilized to investigate and evaluate the effects of patient habitus, acquisition mode, radiation dose, and iterative reconstruction on iodine quantification.

Methods

CT Phantom

A multi-energy CT phantom (Multi-energy CT, Sun Nuclear) (Figure 1) was used with and without an extension ring to simulate two patient sizes (20-cm diameter (‘small’) and 30 × 40 cm (‘large’)). The phantom contained varying concentrations of iodine (0.2, 0.5, 1, 2, 5, 10, and15 mg/ml), blood with iodine (blood + 2 mg/ml iodine and blood + 4 mg/ml iodine), blood, and brain tissue-equivalent inserts. In this study, analysis of iodine measurements was limited to 0.2 - 15 mg/ml iodine inserts. Ground truth iodine densities were based on information provided by the manufacturer. The arrangement of iodine inserts in the phantom is displayed in Figure 1.

• Download figure
• Open in new tab

Figure 1.

Experimental setup. (A, B) Photography of NAEOTOM Alpha with multi-energy CT phantom in large and small configurations. (C, D) Reconstructed iodine map slice of large and small phantom configurations with numbered iodine inserts of varying concentrations: 1, iodine 2 mg/ml; 2, iodine 0.2 mg/ml; 3, iodine 0.5 mg/ml; 4, iodine 5 mg/ml; 5, iodine 1 mg/ml; 6, iodine 10 mg/ml; 7, iodine 15 mg/ml. Iodine density maps are obtained from a single source 120 kVp acquisition mode at 10 mGy and reconstructed using QIR level 4. (WW,WL) = (22, 10) mg/ml.

Results

Measured iodine concentrations and corresponding RMSE values are summarized in Figure 2. Overall, there was consistent iodine quantification accuracy across all iodine inserts between measured and reference values. For iodine concentrations 0.2 - 2 mg/ml, the RMSE values were below 0.3 mg/ml. RMSE values of less than 0.7 mg/ml were observed for 0.5 and 2 mg/ml iodine inserts for large phantom obtained using DS120 and DS140 tube combinations at ultra-low radiation dose levels. The majority of the RMSE values for 5 mg/ml were below 0.4 mg/ml. For 10 mg/ml iodine, the large phantom showed higher RMSE values ranging from 0.3 - 0.7 mg/ml compared to 0.1 - 0.3 mg/ml for the small phantom. The 15 mg/ml iodine insert demonstrated RMSE values less than 0.8 mg/ml except for an outlier of less than 1.15 mg/ml noted for the large phantom obtained using DS120 at ultra-low radiation dose levels.

• Download figure
• Open in new tab

Figure 2.

Comparison of (A) measured iodine densities [mg/ml] and (B) corresponding RMSEs [mg/ml] for individual inserts, ordered by increasing iodine concentration (0.2 mg/ml – 15 mg/ml). Within each column, measurements are ordered by increasing CTDI_vol groups (ultra-low: 0.6, 0.8, 1.2 mGy, low: 1.6, 2, 4 mGy, and medium: 6, 10, 15 mGy)). Phantom size (small / large) and acquisition parameters (SS120, SS140, DS120, and DS140) are coded in different colors and marker symbols.

Figures 3 and 4 illustrate the Bland-Altman plots for low and high iodine concentrations, respectively, separated for each phantom size (small / large) and tube combinations (SS120, SS140, DS120, and DS140) over low radiation dose levels (CTDI_vol = 1.6, 2, 4 mGy). In Figure 3, the mean bias measured for large and small phantom configurations at various tube combinations ranged from -0.11 mg/ml - -0.04 mg/ml and -0.22 mg/ml - -0.14 mg/ml, respectively. In Figure 4, the corresponding mean bias ranges were -0.11 mg/ml - 0.32 mg/ml and -0.44 mg/ml - -0.22 mg/ml for large and small phantoms, respectively. However, only a small difference between individual mean biases was observed across acquisition parameters for both phantom configurations. All bias measurements were within the calculated limits of agreement (LOA). For low iodine concentrations (Figure 3), the widths of the LOA from different acquisition modes ranged from 0.24 mg/ml - 0.44 mg/ml and 0.13 mg/ml - 0.23 mg/ml for the large and small phantoms, respectively. For high iodine concentrations (Figure 4), the corresponding LOA widths ranged from 0.94 mg/ml - 1.34 mg/ml and 0.8 mg/ml - 0.95 mg/ml. Bland-Altman results over ultra-low and medium radiation dose levels are summarized in Tables 2-3 and Tables 4-5, respectively.

• Download figure
• Open in new tab

Figure 3.

Bland-Altman plots illustrating the measured bias [mg/ml] and mean bias [mg/ml] between measured and nominal iodine concentrations versus the average of measured and nominal iodine concentrations [mg/ml] acquired for low iodine concentrations (0.2, 0.5, 1, 2 mg/ml), phantom sizes (small / large), and tube combinations (SS120, SS140, DS120, and DS140) over low radiation dose levels (CTDI_vol = 1.6, 2, 4 mGy). The black dashed line represents the mean bias. The red and blue dashed lines represent the upper and lower limits of the mean bias (from ±1.96σ).

• Download figure
• Open in new tab

Figure 4.

Bland-Altman plots illustrating the measured bias [mg/ml] and mean bias [mg/ml] between measured and nominal iodine concentrations versus the average of measured and nominal iodine concentrations [mg/ml] acquired for high iodine concentrations (5, 10, 15 mg/ml), phantom sizes (small / large), and tube combinations (SS120, SS140, DS120, and DS140) over low radiation dose levels (CTDI_vol = 1.6, 2, 4 mGy). The black dashed line represents the mean bias The red and blue dashed lines represent the upper and lower limits of the mean bias (from ±1.96σ).

The highest mean bias (−0.44 mg/ml) for ultra-low radiation dose levels occurred for the combination of the large phantom, DS120, and high iodine concentration. In Table 2, the mean bias ranges for low iodine concentrations were -0.34 mg/ml to -0.05 mg/ml and -0.16 mg/ml to - 0.06 mg/ml for large and small phantoms, respectively. In Table 3, the mean bias ranges for large and small phantoms were -0.44 mg/ml - 0.21 mg/ml and -0.35 mg/ml to -0.04 mg/ml, respectively. The difference in the mean bias measurements between individual acquisition combinations was minimal and all bias measurements were within the LOA. In Table 2, the widths of the LOA for low iodine concentrations were 0.32 mg/ml - 0.51 mg/ml and 0.12 mg/ml - 0.26 mg/ml for the large and small phantoms, respectively.

For medium radiation dose levels, iodine mean bias measurements ranged from -0.22 mg/ml to -0.12 mg/ml and -0.18 mg/ml to -0.12 mg/ml for large and small phantoms, respectively, considering low iodine concentrations (Table 4). In Table 5, the mean bias measurements ranged from -0.2 mg/ml - 0.31 mg/ml and -0.39 mg/ml to -0.26 mg/ml for the large and small phantoms, respectively. Only minor differences were observed between individual acquisition combinations, phantom configurations, and iodine concentration groups. Again, all bias measurements were within the designated LOA. The widths of the LOA for low iodine concentrations (Table 4) were marked from 0.19 mg/ml - 0.37 mg/ml and 0.04 mg/ml - 0.06 mg/ml for the large and small phantom, respectively. The corresponding widths of LOA for high iodine concentrations (Table 5) were noted from 1.0 mg/ml - 1.12 mg/ml and 0.69 mg/ml - 0.72 mg/ml.

Figure 5 presents the effect of QIR across all radiation dose levels (CTDI_vol = 0.6 - 15 mGy) acquired using SS120 acquisition mode. RMSE values calculated over all (0.2 - 15 mg/ml), low (0.2, 0.5, 1, 2 mg/ml), and high (5, 10, 15 mg/ml) iodine concentrations are shown in three plots. Comparable RMSE values were calculated between individual iterative reconstruction levels across all radiation dose levels in all three cases. On average, overall RMSE values and those for low iodine concentrations were less than 0.3 mg/ml and 0.2 mg/ml, respectively. Although higher RMSE values were calculated for high iodine concentrations, the RMSEs were less than 0.4 mg/ml.

• Download figure
• Open in new tab

Figure 5.

Comparison of RMSE [mg/ml] from different QIR levels (QIR0, QIR2, and QIR4) calculated over (left) all (0.2 - 15 mg/ml), (middle) low (0.2, 0.5, 1, 2 mg/ml), and (right) high iodine concentrations (5, 10, 15 mg/ml), across all radiation dose levels (CTDI_vol = 0.6 - 15 mGy) and obtained using SS120 acquisition mode. In each CTDI_vol column, QIR levels were color-coded and arranged by increasing QIR levels (QIR0, QIR2, and QIR4).

Figure 6 represents the influence of phantom configuration across all radiation dose levels (CTDI_vol = 0.6 - 15 mGy) acquired using SS120 acquisition mode. Computed RMSE values using all (0.2 - 15 mg/ml) and low (0.2, 0.5, 1, 2 mg/ml) were below 0.4 mg/ml while RMSE for high (5, 10, 15 mg/ml) iodine concentrations were less than 0.6 mg/ml. Minimal RMSE variations were evident from CTDI_vol = 1.2 - 6 and 15 mGy over all iodine concentrations. RMSE values calculated over high concentrations showed slight variations across all radiation dose levels except at CTDI_vol = 0.8 mGy. Finally, RMSE values for the small phantom improved with increasing CTDI_vol > 4 mGy.

• Download figure
• Open in new tab

Figure 6.

Comparison of RMSE [mg/ml] from different phantom sizes (small / large) calculated over (left) all (0.2 - 15 mg/ml), (middle) low (0.2, 0.5, 1, 2 mg/ml), and (right) high iodine concentrations (5, 10, 15 mg/ml), across all radiation dose levels (CTDI_vol = 0.6 - 15 mGy) and obtained using SS120 acquisition mode. In each CTDI_vol column, phantom sizes were color-coded.

Discussion

In this study, we present an initial systematic evaluation of a recently introduced clinical DS-PCCT system for measurement of iodine concentrations. We investigated the influence of patient habitus, acquisition parameters, and iterative reconstruction modes on the accuracy of DS-PCCT iodine-specific images or iodine maps. Overall, our phantom study demonstrated accurate iodine quantification with minimal effect of patient habitus, tube voltage combinations, or iterative reconstruction on the iodine measurement accuracy of the DS-PCCT system, including at considerably lower than dose levels than used in the current clinical day-to-day routine. It was noted that despite the minimal increase in iodine bias, PCCT technology outperforms other spectral platforms due to its excellent ultra-low dose performance.

DS-PCCT measurements of various iodine concentrations demonstrated comparable accuracy compared to ground truth over a wide range of values (0.2 to 15 mg/ml). This included high measurement accuracy at low iodine concentrations (0.2 - 2 mg/ml) resulting in RMSE of less than 0.3 mg/ml across different phantom sizes, tube settings, and radiation dose groups. Based on data collected from an early prototype PCCT system, Riederer et al., reported similar iodine quantification performance¹⁶. Reliable quantification of iodine uptake, particularly of low iodine concentrations, may enable improved diagnosis including early tumor detection, differentiation, and staging. On the other hand, high iodine concentrations were associated with a slight increase in RMSE. Lang et al., on a research PCCT system, demonstrated similar observations¹⁷. All biases measured in this study were well within the LOA, though even lower measurement variability was seen with low iodine concentrations, the small phantom, and medium radiation dose levels.

Previous DECT studies reported a significant effect of patient habitus on the accuracy of iodine quantification at ultra-low radiation exposures^18,19. The advent of PCD technology has improved the quality of diagnostic imaging of large patients significantly compared to EIDs^20,21. In our quantitative evaluation, calculated RMSEs over all iodine concentrations for both phantom configurations remained consistently below 0.4 mg/ml, even for very low dose levels. Furthermore, only minimal RMSE discrepancies were seen between phantom configurations. These suggest accurate and consistent iodine concentration measurements regardless of the patient size.

Similar iodine quantification performance was observed across the iterative reconstruction levels at all radiation dose levels. These results indicate that the noise reduction, which is associated with QIR strengths, does not significantly affect iodine measurements. Similar observations have been made in phantom studies on available DECT systems^18,19,22 and clinical studies with the DS-PCCT^12,23.

We used a phantom model instead of patient data for analysis, which has several limitations. In clinical scans, iodine quantification can be affected by factors such as respiratory motion, metal artifacts, or morbid obesity. To overcome some of these limitations, multiple tube settings and phantom sizes were used in this study, but patients studies are still needed to account for these factors.

Conclusion

Quantification of iodine uptake can assist in clinical diagnosis, such as distinguishing between a hyperdense cyst and a solid tumor. Accurate and reliable iodine density maps may eliminate the need for the unenhanced phase, which has served as a comparison for determining iodine uptake. It is anticipated that, in the future, accurate and precise iodine quantification will provide opportunities for the development of spectral-based biomarkers, such as those that may be employed to evaluate cancer treatments.