4.1 Correlation and consistency between the MS and DC methods

In this study, we found that the BV and BF values had no significant differences between different pancreatic regions for either algorithm (Table 1). This was consistent with the previous reports. For instance, Zhu et al.21 compared the BF values of the pancreatic head, body, and tail in 32 subjects with normal pancreas and prove that there are no significant differences in BF values between the different pancreatic regions. Besides, Delrue et al.22 evaluated the normal pancreatic head, body, and tail in 20 subjects and confirmed that BV and BF had no significant differences between the different pancreatic regions.

Furthermore, the results of Pearson linear correlation analysis indicated that there was a positive correlation between BV and BF values of pancreatic head, body, and tail of the two algorithms. Kishimoto23 compared the BV and BF values of nine canine pancreases between the MS and DC methods and confirmed a good correlation between the two methods. This was consistent with the results of the present study, but they do not analyze the consistency of the results between the two methods. Kanda et al.9 compared both the correlation and consistency of the BF of the hepatic artery and portal vein between the MS and DC methods in 88 patients and the results show a good correlation between the two methods, but the values obtained using the MS method are significantly higher than those obtained using the DC methods. Similarly, our study showed that the BV and BF values calculated using the MS method were higher than those calculated using the DC methods. In particular, the BV values obtained using the MS method were three times those obtained using the DC method. The reason might be that the MS algorithm assumed that there was no outflow of contrast agent for a short period before the maximum TDC slope of the tissue was reached after the injection of the contrast agent. However, in the actual physiological state, contrast agent outflow is present. Thus, the perfusion parameters obtained using the MS method are higher than those in the actual physiological state. In contrast, the DC algorithm does not assume the tissue perfusion model and the result is closer to the actual physiological state.9, 24

Bland–Altman statistical analysis was used to evaluate the consistency of the perfusion parameters between the MS and DC methods and the results showed that the mean difference value of BV and BF values in both the subjects with the normal pancreas and those with pancreatitis were not equal to 0, indicating that the values of the perfusion parameters measured by the two mathematical algorithms were not consistent. Besides, within the bounds of the 95% consistency, the difference in the perfusion parameters between the two algorithms (the absolute value of the maximum difference) was very large, which was not acceptable in clinical practice. This highlighted the impact of the mathematical algorithms on the calculation of perfusion parameters. Due to the inconsistency of the perfusion parameters obtained by different algorithms, it should be noted that the thresholds for perfusion parameters based on the DC method could not be applied to the MS method when performing clinical disease analysis and multi-center data exchange research, and vice versa. The results of this study remind us that we should pay attention to the algorithm used when using perfusion parameters to determine the severity of pancreatitis and monitor the efficacy. Because the perfusion parameters obtained by different algorithms are inconsistent, the threshold values of the perfusion parameters obtained by the DC method should not be used to judge the severity of pancreatitis measured by the MS method and monitor the efficacy. The conclusion of this study application had certain significance for clinical practice, the application of perfusion parameters in clinical practice, especially the threshold to differentiate benign tumors and monitoring curative effect should first clear the perfusion parameters adopted by the mathematical algorithm, might not be based on the DC method of perfusion parameters threshold directly to the MS method was used to measure of benign and malignant tumor, curative effect monitoring clinical situations.

4.2 Limitations of the study

First, the locations and sizes of the ROIs might affect the measurement results. Second, in this study, the scanning initiation time was 10 s after contrast injection, and no individual factors such as cardiac function were taken into consideration. However, the MS method was more affected by cardiac parameters (such as cardiac output) compared with the DC method.24 Third, the high flow rate of the contrast agent might affect the results. Therefore, they were not transferable to routine clinical practice.