This retrospective study was approved by the Institutional Review Board of the Peking University School of Oncology, and written informed consent was waived. From January 2016 to July 2018, a total of 1468 consecutive patients with focal liver lesions were referred to our department for CEUS examinations. Of the 109 hemangioma cases, 23 cases had atypical CEUS patterns and previous CRC histories, which were difficult to exclude from liver metastasis. During the same period, 480 patients with suspected liver metastasis from CRCs were diagnosed based on CEUS. Because there were much more liver metastasis cases (n = 480) than hemangioma cases (n = 20), we randomly selected 24 cases from the liver metastasis pool according to a 1:20 proportion. If patients had more than one lesion, the largest and most clearly presented lesion was chosen for CPI evaluation. Five patients who had deep breath during the arterial phase and poor imaging quality were excluded. Finally, 20 hemangiomas and 22 liver metastases were entered into this retrospective study (Figure 1).

Open in New Tab   Full Size Figure   Download Figure

Figure 1 Flow diagram of the study population. CPI: Color parametric imaging; FLL: Focal liver lesions, CEUS: Contrast enhanced ultrasound.

According to the liver CEUS guidelines[14] and other studies[10], the atypical pattern of CEUS for hemangioma included rapid homogeneous hyperenhancement at the arterial phase and hypoenhancement at the portal/late phase and peripheral nodular enhancement at the arterial phase and lack of enhancement in the center at the late phase. The CEUS pattern for liver metastasis included hyperenhancement at the arterial phase and washout at the late phase or rim-like enhancement at the arterial phase and a nonenhancement area at the late phase. Among the 22 patients with liver metastasis, the final diagnosis was confirmed by pathologic analysis of specimens obtained via US-guided percutaneous biopsy (n = 16) or surgical resection (n = 6). Among the 20 hemangiomas, the final diagnosis was based on either pathological results (n = 9) or contrast-enhanced computed tomography or contrast-enhanced magnetic resonance imaging findings with at least one year follow-up (n = 11). Among the 42 patients, 19 were male and 23 were female. The average age was 60.5 ± 9.3 years (range: 39-75 years). The mean size of liver lesions was 3.2 ± 1.8 cm (range: 1.3-11 cm).

A Logiq E9 ultrasonic machine (GE Healthcare, Milwaukee, WI, United States) was used with a C1-5 convex probe to obtain routine US and CEUS images. The ultrasound contrast agent was SonoVue (Bracco, Milan, Italy). Lyophilized SonoVue powder was dissolved in 5 mL saline. Bolus injection (2 mL the suspension) was performed at the antecubital vein via a 20G cannula within 2-3 s, followed by a 5-mL saline flush.

Before the examination, the patients were required to lie in the left lateral position or supine position and breathe steadily. The liver lesions were scanned and located using routine US. The echogenicity, diameter, border, morphology, necrosis, halo sign and vessels were observed. It was defined as necrosis in ultrasound imaging if there was an anechoic area within the lesions, clear boundaries, and color Doppler flow imaging showed no blood flow within the anechoic area. The halo sign was defined as a hypoechoic rim found around the solid mass with a distinct difference between the lesion and the surrounding liver. At least two vessels inside the lesion on color Doppler flow imaging indicated rich flow.

Then, the contrast mode was entered. The imaging settings, such as gain, depth, and focus, were optimal. The mechanical index was set at 0.11-0.13. After injecting contrast agent, the liver was scanned using contrast-enhanced harmonic grayscale sonography, and timer was initiated simultaneously. The dynamic blood perfusion of lesions was observed from baseline to the late phase. Consecutive cine clips (90 s each) were recorded and stored on the hard disk for further analysis. The enhancement phases were divided into the arterial phase (10-30 s), portal vein phase (31-120 s), and late phase (120-360 s) after injection of contrast agent. The same ultrasound machine with the procedure software of parametric imaging was used to obtain CPI.

Two resident radiologists (W. X. F. and W. H.) each with at least one year of experience in the evaluation of liver CEUS images as well as two staff radiologists (Y. W. and W. W.) each with at least 10 years of experience in the liver CEUS images retrospectively read all the CEUS and CPI images independently. Two radiologists at the same level retrospectively interpreted the CEUS images without knowledge of the patients’ final diagnosis. In all cases, consensus agreement between the two radiologists was used to determine the CEUS feature and possible diagnosis. According to CEUS guidelines[14], the enhancement patterns at the arterial phase to portal phase of liver lesions included four patterns: Peripheral nodular enhancement (peripheral focal enhancement in the arterial phase, progressing in a centripetal direction to partial fill-in.), rim-like enhancement (peripheral enhancement in the arterial phase, without gradual fill-in.), homogenous hyperenhancement, and heterogeneous enhancement.

All of the CEUS images of the arterial phase were reconstructed with the Parametric Imaging program using a Logiq E9 XD Clear ultrasonic machine (GE Healthcare, Milwaukee, WI, United States). The CEUS video clips were reviewed to record the starting and ending points of lesion enhancement. In the CPI system, time zero was regarded as the point at which the contrast agent reached the liver, and the arrival time was then calculated between the current time and time zero. If the time zero was set at the time when contrast agent was injected, the arrival time of contrast agent at each pixel could be calculated. Thus, the color map consisted of individual pixels representing the arrival time of the contrast agent in the tumor. The residents and staff radiologists read all the CPI images blindly and independently. The CPI enhancement patterns of liver lesions were summarized and classified into four patterns: (1) Peripheral nodular enhancement (round or semicircle shaped enhancement visualized at the peripheral area of lesions and no enhancement in the center); (2) Peripheral rim-like enhancement (peripheral enhancement without nodular or regular shape); (3) Concentric circle enhancement (multiple circles with the same center); and (4) Mosaic enhancement (enhancing lines of the vascular tree, defined as one or more hypertrophic tortuous arteries that reached the edge of the lesion, partially encircling the nodule and penetrating internally with a basket or chaotic distribution) (Figure 2). The feeding artery, defined as a hypertrophic artery that was directed toward the lesion and was larger than the branches at the same depth during the arterial phase, was red in the central area, which was surrounded by yellow, or was a scattered distribution of red and yellow (Figure 2). In all cases, consensus agreement between the two radiologists was used to determine the CPI feature and possible diagnosis.

Open in New Tab   Full Size Figure   Download Figure

Figure 2 color parametric imaging patterns of liver atypical hemangioma and liver metastases. First line was sketch figures for the four enhancement patterns of color parametric imaging. Second line was representative routine contrast-enhanced ultrasound images corresponding to the four enhancement patterns. Third line was representative color parametric images corresponding to the four patterns. A: Peripheral nodular enhancement; B: Peripheral rim-like with feeding artery (▲); C: Concentric circles enhancement; D: Mosaic enhancement with feeding artery (▲).

In the training, both staff and resident radiologists were required to review all 20 cases of hemangioma and liver metastasis with typical CEUS and CPI features. CEUS and CPI diagnoses were scored using a 5-point scale: 1 = hemangioma with strong evidence, 2 = possible hemangioma, 3 = undetermined, 4 = possible liver metastasis, and 5= liver metastasis with strong evidence.

The arrival time of the contrast agent after injection could be displayed at any point of CPI. The arrival time of a lesion was defined as the arrival time of the earliest color point in the lesion, and the peak time was regarded as the arrival time of the last color point in the lesion. The peak time of CPI was regarded as the time point where the brightest color was in the lesion. △AT was regarded as the difference in peak time and arrival time of CPI of atypical hemangioma or liver metastasis.

SPSS 21.0 statistical software was used for statistical analysis. The quantitative data are displayed as the mean ± SD and were compared by t test. The Kappa test was used to analyze the interrater agreement between the staff and resident radiologists. The agreement was graded as follows: Moderate (0.2-0.39), fair (0.40-0.59), good (0.60-0.79), and perfect (0.80-1.0) agreement. The diagnostic sensitivity, specificity, and accuracy between CEUS and CPI patterns were compared by the McNemar test. The count data were analyzed using χ² test and Fisher’s exact test. A P value less than 0.05 was considered statistically significant.

The routine CEUS features of liver atypical hemangioma and liver metastases analyzed by staff and resident radiologists are shown in Table 1.

The CEUS features of peripheral nodular enhancement were observed during the arterial phase more frequently in patients with atypical hemangioma than in those with metastasis (P = 0.003) by staff radiologists. The CEUS features of heterogeneous hyperenhancement were observed during the arterial phase significantly more frequently in patients with metastasis than in those with atypical hemangioma by staff radiologists (P = 0.023). However, the perfusion patterns detected by resident radiologists were not as sensitive as those detected by staff radiologists (Table 1). The features of peripheral nodular enhancement and heterogeneous hyperenhancement for resident radiologists were not significantly different between atypical hemangioma and metastasis (P = 0.052, P = 0.096). The feeding arteries for the staff and resident radiologists were at a power of 25%–45.5%.

The CPI features of liver atypical hemangioma and liver metastases analyzed by staff and resident radiologists are shown in Table 2.

The CPI features of peripheral nodules were observed during the arterial phase more frequently in patients with atypical hemangioma than in those with metastasis by both groups of radiologists (65%-70.0% vs 4.5%-13.6%, P < 0.001, P = 0.001). In addition, the CPI features of mosaic enhancement (5%-10% vs 68.2%-63.6%, P < 0.001, P < 0.001) and feeding artery (20% vs 59.1%-54.5%, P = 0.010, P = 0.021) were found during the arterial phase more frequently in patients with metastasis than in those with atypical hemangioma by both groups of radiologists. The feeding arteries for the staff and resident radiologists were at a power of 20%–59.1%. CPI imaging offered significant improvements in detection rates compared with routine CEUS signs in both groups.

The consistency of CEUS and CPI between staff and resident radiologists was analyzed (Table 3). With regard to CEUS features found by staff and resident radiologists, the diagnostic consistency in homogenous hyperenhancement and heterogeneous enhancement was perfect (k = 0.835, k = 0.804), and the diagnostic consistency in peripheral nodular enhancement and rim-like enhancement was good (k = 0.602, k = 0.692).

With regard to CPI features found by staff and resident radiologists, the diagnostic consistency in peripheral nodular enhancement and mosaic enhancement was perfect (k = 0.885, k = 0.803), and the diagnostic consistency in peripheral rim-like enhancement and concentric circle enhancement was good (k = 0.713, k = 0.760).

The diagnostic confidence scoring of lesions by staff and resident radiologists is summarized in Figure 3. There were significant differences in the final score distribution between the two methods. The number of 3-score (undetermined diagnosis) in CEUS was obviously higher than that in CPI by both the groups of staff (45.2% vs 4.8%, P < 0.001) and resident radiologists (35.7% vs 7.1%, P = 0.001).

Open in New Tab   Full Size Figure   Download Figure

Figure 3 Diagnostic confidence of atypical hemangioma and liver metastasis by staff and resident radiologists. The number of 3-score (undetermined diagnosis) in contrast enhancement ultrasound was significantly higher than that in color parametric imaging in both staff group (A) and resident group (B). CEUS: Contrast enhancement ultrasound; CPI: Color parametric imaging.

The diagnostic performance of CEUS and CPI by staff and resident radiologists is shown in Table 4.

By resident radiologists, the specificity of CEUS+CPI was significantly increased compared with that of CEUS (77.3% vs 45.5%, P = 0.030). The sensitivity of CEUS+CPI was higher than that of CEUS, but the differences were not significant (80.0% vs 55.0%, P = 0.456). The accuracy of CEUS+CPI was significantly higher than the accuracy of CEUS (78.6% vs 50%, P = 0.006). Additionally, the AUC of CEUS+CPI was significantly higher than that of CEUS (AUC = 0.803 vs AUC = 0.757, P = 0.036).

By staff radiologists, accuracy was improved in CEUS+CPI (81.0% vs 54.8%, P = 0.010), whereas no significant differences in specificity and sensitivity were found (P = 0.144, P = 0.112). The AUC of CEUS+CPI was significantly higher than the AUC of CEUS (0.890 vs 0.825, P = 0.013) by staff radiologists.

The comparison of the △AT of CPI between atypical hemangioma and liver metastasis is summarized in Figure 4. The difference in peak time and arrival time of CPI of atypical hemangioma was significantly longer than that of liver metastasis (8.31 ± 3.05 s vs 5.13 ± 0.99 s, P < 0.001).

Open in New Tab   Full Size Figure   Download Figure

Figure 4 Spot diagram of △AT in atypical hemangioma and liver metastasis. △AT = Difference valve of peak time and arrival time of color parametric imaging of liver lesions. △AT of atypical hemangioma was significantly longer than that of liver metastasis (8.31 ± 3.05 s vs 5.13 ± 0.99 s, P < 0.001).

In clinical practice, the diagnosis is sometimes difficult with contrast-enhanced ultrasound (CEUS) when the case has an atypical perfusion pattern. Color parametric imaging (CPI) is an analysis software for CEUS with better detection of temporal differences in CEUS imaging using arbitrary colors. It measures the differences in arrival time of the contrast agent in lesions so that the perfusion features of atypical hemangioma and colorectal cancer liver metastasis can be distinguished.

The motivation of this study was to evaluate the role of CPI in the differential diagnosis of atypical hemangioma from liver metastases and the diagnostic performance by staff and resident radiologists. The patients with atypical hemangioma would be benefited by avoiding invasive test or even surgical resection. Furthermore, a junior radiologist can be more confident in the differential diagnosis of liver lesions by CPI.

To evaluate the role of a novel type of CPI of CEUS in the differential diagnosis of atypical hemangioma from liver metastases in patients with a history of colorectal cancer.

All enrolled patients received ultrasound, CEUS and CPI examinations. Resident and staff radiologists independently and retrospectively reviewed CEUS and CPI images. Two sets of criteria were assigned: (1) Routine CEUS alone; and (2) CEUS and CPI. The diagnostic sensitivity, specificity, accuracy and receiver operating characteristic (ROC) curve of resident and staff radiologists were analyzed.

The following CPI features were significantly different between liver hemangioma and liver metastases analyzed by staff and resident radiologists: Peripheral nodular enhancement (65%-70.0% vs 4.5%-13.6%, P < 0.001, P = 0.001), mosaic/chaotic enhancement (5%-10% vs 68.2%-63.6%, P < 0.001, P < 0.001) and feeding artery (20% vs 59.1-54.5%, P = 0.010, P = 0.021). CPI imaging offered significant improvements in detection rates compared with routine CEUS signs in both resident and staff groups.

CPI could provide specific information on the hemodynamic features of liver lesions and help to differentiate atypical hemangioma from liver metastases, for both staff and resident radiologists. CPI is useful especially for radiologists with less CEUS experience.

In this study, a novel type of color contrast enhanced ultrasound provided supplemental information for differential diagnosis between atypical hemangioma and liver metastasis. This technique is safe and effective in clinical practice. However, to confirm the performance of this new imaging method, studies on a larger sample set are required.