A total of 52 confirmed CRLM in 17 patients were included in the study.
CRLM size ranged from 0.3 - 9 cm (mean 2.7 cm).
CRLM in all hepatic segments were represented.
Eight patients had a solitary CRLM,
while 9 patients had 4 - 6 CRLM.
Individual sensitivity,
specificity and AUC values for each of the 6 readers for both the "standard protocol" sequences and "standard + Primovist protocol" sequences are shown in figures 1 and 2,
respectively.
Individual ROC curves for each of the 6 readers for both the "standard protocol" sequences and "standard + Primovist protocol" sequences are shown in figures 3 and 4,
respectively.
The individual results demonstrate that 2 readers performed better with the “standard protocol” sequences (Junior Staff radiologist and 4th year radiology resident),
2 readers performed better with the “standard + gadoxetic acid protocol” sequences (Senior Staff radiologist and 4th year radiology resident) and 2 readers performed the same with both (Senior Staff radiologist and Junior Staff radiologist).
Combined sensitivity,
specificity and AUC values for all 6 readers for both the "standard protocol" sequences and "standard + Primovist protocol" sequences are shown in figure 5.
Combined ROC curves for all 6 readers for both the "standard protocol" sequences and "standard + Primovist protocol" sequences are shown in figure 6.
The combined results demonstrate no statistical difference between the 2 protocol sequences (p = 0.91).
Therefore,
ROC analysis shows no statistical difference between "standard protocol" sequences and "standard + gadoxetic acid protocol" sequences in lesion detection and declaring segments free of disease.
It is not yet clear whether this is clinically significant (ie.
changes diagnostic thinking or management on a per-patient basis).
Examples of cases are as follows:
Case 1 illustrates concordance between the portal venous (PV) phase and Primovist-specific sequences (Figures 7 and 8).
Case 2 illustrates the complementary nature of the PV phase and Primovist-specific sequences (Figures 9 and 10).
Cases 3 - 5 illustrate improved lesion conspicuity of the Primovist-specific sequence over the PV phase (Figures 11 - 16).
Cases 6 and 7 illustrate increased lesion detection by the Primovist-specific sequence over the PV phase (Figures 17 - 20).