CEUS-based quantitative analysis of washout time in focal liver lesions affects hepatocellular carcinoma diagnosis

Introduction

Primary liver cancer (PLC) is the sixth most common malignancy worldwide and the third leading cause of cancer-related death (1). Hepatocellular carcinoma (HCC), the most prevalent subtype of PLC, accounts for approximately 75–85% of cases (2). In recent years, real-time low mechanical index contrast-enhanced ultrasound (CEUS) has been established as a useful tool in the multimodal diagnostic approach to characterize hepatic nodules (3). The American Association for the Study of Liver Diseases (AASLD) 2023 guidelines emphasize that CEUS is recognized as a valuable complementary imaging modality for further characterizing liver lesions that cannot be definitively diagnosed by contrast-enhanced computed tomography (CECT) or contrast-enhanced magnetic resonance imaging (CEMRI) (4). The main advantage of CEUS over other imaging modalities is its high temporal resolution, enabling real-time evaluation of the liver. Moreover, CEUS offers the highest contrast resolution of any clinical imaging modality (5). It has been reported that, compared with CECT and CEMRI, CEUS serves as an effective diagnostic tool for HCC, with sensitivity and specificity ranging from 69–73% and 88–95%, respectively (6-8). Given its widespread clinical use, CEUS was incorporated into the American College of Radiology Liver Imaging Reporting and Data System (LI-RADS) in 2016, with subsequent updates released in 2017 (9).

Washout refers to the imaging feature in which the enhancement of a lesion becomes partially or completely lower than that of the surrounding liver parenchyma, making the lesion appear darker (lower grayscale) than the adjacent tissue. Washout time, which is defined as the timing when a lesion becomes visibly less enhanced than the surrounding liver parenchyma after intravenous injection of the contrast agent, plays a pivotal role in lesion characterization within the CEUS LI-RADS (version 2017) (10,11). According to CEUS LI-RADS, lesions exhibiting washout within 60 seconds are considered to exhibit early washout and are classified as LR-M (defined as probably or definitely malignant but not HCC-specific). This 60-second criterion is uniquely important in CEUS LI-RADS because microbubble contrast agents remain strictly intravascular, enabling more reliable assessment of washout. In contrast, CECT and CEMRI contrast agents diffuse into the extracellular interstitial space and may obscure washout in some malignancies, particularly in non-HCC malignancies, as well as a proportion of nontypical HCC (12). In the study by Zheng et al., which included 2,020 liver nodules, HCC accounted for 63% (224/354) of LR-M lesions, with 96% (214/224) of these HCCs classified as LR-M due to early washout within 60 seconds (13). In our previous investigation on small HCC, the proportion of HCC within the CEUS LR-M category was as high as 75% (14). LR-5 was defined by CEUS LI-RADS (2017) as the category indicating liver lesions that are definitely HCC. By applying 45 seconds as the early washout threshold for reclassification, Ding et al. reported improved sensitivity of LR-5 for HCC (76.2% vs. 65.5%; P=0.012), a higher area under the receiver operating characteristic curve (AUC) (0.85 vs. 0.80; P=0.001), and increased specificity of LR-M for diagnosing non-HCC malignancies (81.3% vs. 71.4%; P=0.010) (15). Furthermore, in a very recent study, Wang et al. demonstrated that arterial-washout temporal interval (AWTI), defined as the time between arterial phase onset and washout onset, reduced HCC misclassification and significantly improved the diagnostic performance of CEUS LI-RADS for both LR-M and LR-5 when used instead of conventional washout onset time alone (16).

At present, the evaluation of early washout is primarily based on visual analysis (VA). In the study by Schellhaas et al., the interobserver agreement for washout assessment was reported to be moderate (κ=0.49) (17). Zhou et al. suggested that the subjective perception of visual washout on CEUS constitutes a major limitation of CEUS LI-RADS, highlighting the need for further research to mitigate subjectivity and enhance interobserver agreement (18). Against this background, quantitative analysis (QA) using dedicated CEUS processing software—with the potential for future integration with artificial intelligence (AI)—is showing considerable promise as a new evaluation method. In current clinical practice, QA is employed to evaluate real-time dynamic images, from which time-intensity curves (TICs) are generated (19). From these TICs, several key quantitative parameters can be extracted, including washout time, peak intensity, and time to peak (20,21). This approach offers the potential to overcome the inherent subjectivity and variability of VA.

However, it remains unclear what discrepancies exist between QA and VA regarding washout time, and how these differences affect the CEUS LI-RADS classifications. To address this gap, the present study was designed to compare washout times measured by QA and VA, and to examine their consistency in CEUS LI-RADS classifications. Furthermore, we explored an optimal QA-based washout threshold aimed at improving the diagnostic efficacy of QA LR-5 for HCC. This work serves as an initial step toward the development of standardized, objective criteria that could facilitate future AI-assisted diagnostic frameworks. We present this article in accordance with the STARD reporting checklist (available at https://qims.amegroups.com/article/view/10.21037/qims-2026-1-0144/rc).

Methods

Study population

This study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. This retrospective study was approved by the Ethics Committee of West China Hospital, Sichuan University (No. 2020-708). The collection of informed consent from patients was deemed unnecessary and waived due to the retrospective nature of the study. This study was conducted at the Department of Medical Ultrasound, West China Hospital, Sichuan University. Patients who underwent liver CEUS examinations between April 2020 and October 2021 were enrolled at our hospital. Clinical information was retrieved from the hospital information system, including patient gender, age, high-risk factors for HCC (such as chronic hepatitis B virus infection, liver cirrhosis, or a prior diagnosis of HCC), and the pathological diagnoses of focal liver lesions (FLLs). The inclusion and exclusion criteria were defined as follows:

• Inclusion criteria: (I) patients with definite FLLs identified by baseline ultrasound (US) who underwent liver CEUS examinations; (II) FLLs with confirmed pathological diagnoses; (III) patients at risk for HCC; (IV) patients with sufficient CEUS imaging data and clinical information.
• Exclusion criteria: (I) patients with poor CEUS imaging quality; (II) patients with history of treatment for FLLs, including ablation, transarterial chemoembolization, or immunotherapy.

A total of 233 FLLs from 225 patients were ultimately included in the final analysis. The detailed patient selection and comparison strategy of this study are illustrated in Figure 1.

Figure 1 Flowchart of study population selection and comparison strategy framework. The study enrolled patients at high risk for HCC and evaluated their FLLs according to predefined inclusion and exclusion criteria. A total of 233 FLLs from 225 patients were included in the final analysis. These lesions were evaluated using both QA and VA to compare washout times, the consistency in CEUS LI-RADS classifications, and the diagnostic performance for HCC. CEUS, contrast-enhanced ultrasound; FLLs, focal liver lesions; HBV, hepatitis B virus; HCC, hepatocellular carcinoma; LI-RADS, Liver Imaging Reporting and Data System; QA, quantitative analysis; US, ultrasound; VA, visual analysis.

US examination

All enrolled patients underwent baseline US and CEUS examinations using the Resona 7 system (Mindray, China). After conventional liver US evaluation, the lesion characteristics—including size, location, echotexture, echogenicity, and margins—were documented. CEUS was subsequently performed at a low mechanical index (mechanical index <0.1). Following intravenous injection of 1.2–2.4 mL of contrast agent (SonoVue, Bracco, Italy), the CEUS timer was immediately activated, and a 5 mL saline flush was administered to clear the catheter. To reduce microbubble destruction from US exposure, continuous imaging of the target lesion and adjacent parenchyma was performed during the initial 60 s, followed by intermittent scanning until either definite washout was observed or parenchymal enhancement subsided. All imaging data were stored for subsequent evaluation.

Reference standard and diagnostic standard

Histopathological diagnosis obtained through surgical resection or US-guided biopsy served as the reference standard in this study. Lesions classified as LR-5 by either VA or QA on CEUS were considered diagnostic for HCC (14).

CEUS VA

Clinical information and pathological results were anonymized. Two additional radiologists, with over 5 and 3 years of CEUS experience respectively, independently reviewed all CEUS examinations of FLLs. To minimize interpretation bias, the readers were blinded to the pathological reference standard and to the QA results during image review. They visually assessed the presence of contrast agent washout and recorded the corresponding washout time. Lesions were then classified according to CEUS LI-RADS (version 2017) as follows: LR-1 (definitely benign), LR-2 (probably benign), LR-3 (intermediate probability of malignancy), LR-4 (probably HCC), LR-5, and LR-M (22). Interobserver agreement was assessed, and discrepancies were resolved by joint review to reach consensus, which was used for final classification.

CEUS QA

CEUS dynamic images were processed using VueBox^® (Bracco, Italy), professional software referenced in the 2023 technical review by the European Federation of Societies for Ultrasound in Medicine and Biology (EFSUMB) (19). Three regions of interest (ROIs) were placed: the first ROI defined the overall analysis area to exclude non-echogenic data, while two additional ROIs were placed within this area to outline the lesion and the adjacent liver parenchyma at the same depth as the lesion (23). Large vessels and necrotic areas were carefully avoided. Automatic motion correction was applied to minimize respiratory motion artifacts. The software then generated TICs for both the lesion and surrounding parenchyma (Figure S1), with the x-axis representing time and the y-axis indicating echo signal intensity. Washout time was defined as the intersection of the lesion and parenchyma TICs, occurring when the lesion’s signal intensity changed from higher to lower than that of the parenchyma (11,24). Curve-fitting results were considered reliable when the fitting quality (agreement between raw data and theoretical curve) exceeded 75% (25). Lesions classified as LR-5 by QA according to CEUS LI-RADS (version 2017) were designated as “QA LR-5” to distinguish them from LR-5 lesions identified by VA (designated as “VA LR-5”). Given the continuous nature of washout, multiple candidate thresholds (60, 55, 50, 45, 40, 35, and 30 seconds) were tested to identify the optimal cutoff, which was then used to define “modified QA LR-5”. QA was independently performed by a specialized radiologist in CEUS. Except for the modified QA LR-5 criteria, all other classifications strictly adhered to CEUS LI-RADS version 2017.

Statistical analysis

Quantitative data were presented as mean ± standard deviation, while qualitative data as counts with corresponding percentages. Each FLL was treated as a single unit of analysis, rather than each patient. Wilcoxon signed-rank tests were used to compare median washout times of FLLs and HCCs between VA and QA. Weighted κ values were calculated to assess interobserver agreement in CEUS LI-RADS classifications between radiologists, as well as agreement between VA- and QA-based classifications. Agreement was interpreted as follows: κ value of ≤0.2, poor; 0.21–0.4, fair; 0.41–0.60, moderate; 0.61–0.8, good, and 0.81–1.0, almost perfect. VA LR-5, QA LR-5 and modified QA LR-5 were considered diagnostic for HCC. Their sensitivity, specificity, accuracy were compared using the McNemar test, while the AUC was compared using the DeLong test. Statistical analyses were performed using MedCalc software (version 10.4.7.0, Ostend, Belgium), and a P<0.05 was considered statistically significant.

Results

Patient and nodule characteristics

This study included 233 FLLs from 225 patients, including 8 patients with two FLLs each. Among the 225 patients (mean age, 57.0±10.6 years), 170 (75.6%) were men. The mean size of liver nodules was 3.3±1.7 cm. Detailed clinical characteristics of the patients are presented in Table 1. In our study, all FLLs were pathologically confirmed, with 215 lesions (92.3%) obtained via surgical resection and 18 lesions (7.7%) via US-guided biopsy. The pathological composition of tumors is also displayed in Table 1, comprising 220 (94.4%) malignant lesions including 184 (79.0%) HCCs (Figure 2) and 36 (15.4%) non-HCC malignancies (Figure 3), as well as 13 (5.6%) benign lesions (Figure 4).

Figure 2 A 61-year-old man with pathologically confirmed well-differentiated HCC. (A) A hypoechoic lesion (arrow) measuring 1.2 cm × 0.9 cm in the left lateral lobe of the liver was detected on conventional US. (B) Homogeneous hyperenhancement (arrows) in the arterial phase was shown. (C) The lesion exhibited isoenhancement (arrows) in the portal phase. (D) Late mild washout (arrows) was shown. (E) Dual-ROIs were placed with the lesion delineated by a green contour and the adjacent hepatic parenchyma marked by an orange circle, and the blue circle indicaes the overall analysis ROI used to exclude non-echogenic data. (F) No early washout was demonstrated on TIC by QA (lesion, green curve; adjacent hepatic parenchyma, orange curve). The lesion was classified as LR-5 by both VA and QA. HCC, hepatocellular carcinoma; LR-5, defined by CEUS LI-RADS [2017] as indicating lesions that are definitely HCC; QA, quantitative analysis; ROIs, regions of interest; TIC, time-intensity curve; US, ultrasound; VA, visual analysis.

Figure 3 A 65-year-old woman with pathologically confirmed intrahepatic cholangiocarcinoma. (A) A hypoechoic mass (arrow) measuring 3.6 cm × 3.3 cm in the right lobe of the liver was detected on conventional US. (B) Hyperenhancement (arrow) in the arterial phase was shown. (C) Early washout (arrow) in the portal phase was detected by VA. (D) Mild hypoenhancement (arrow) in the late phase was shown. (E) ROIs were placed with the lesion delineated by a green contour and the adjacent hepatic parenchyma marked by an orange circle, and the blue circle indicates the overall analysis ROI used to exclude non-echogenic data. (F) Early washout around 30 seconds after contrast agent injection was clearly demonstrated by TIC on QA (lesion, green curve; adjacent hepatic parenchyma, orange curve; washout onset at 30 seconds on QA). The lesion was classified as LR-M by both VA and QA. HCC, hepatocellular carcinoma; LR-M, defined by CEUS LI-RADS (2017) as lesions that are probably or definitely malignant but not HCC-specific; QA, quantitative analysis; ROIs, regions of interest; TIC, time-intensity curve; US, ultrasound; VA, visual analysis.

Figure 4 A 54-year-old woman with chronic hepatitis B and pathologically confirmed focal nodular hyperplasia. (A) A slightly hyperechoic mass (arrows) measuring 2.0 cm × 1.5 cm in the left lateral lobe was detected on conventional US. (B) Hyperenhancement (arrow) in the arterial phase was shown at CEUS. (C) No washout was observed (arrow) in the portal phase. (D) Persistent enhancement (arrow) in the late phase was shown at CEUS. (E) ROIs were placed with the lesion delineated by a green contour and the adjacent hepatic parenchyma marked by an orange circle, and the blue circle indicaes the overall analysis ROI used to exclude non-echogenic data. (F) No early washout was demonstrated on TIC by QA (lesion, green curve; adjacent hepatic parenchyma, orange curve). The lesion was classified as LR-4 by both VA and QA. CEUS, contrast-enhanced ultrasound; LR-4, defined by CEUS LI-RADS [2017] as indicating lesions that are probably HCC; QA, quantitative analysis; ROIs, regions of interest; TIC, time-intensity curve; US, ultrasound; VA, visual analysis.

Washout time of HCCs assessed by QA and VA

The washout times for HCCs obtained by QA and VA are compared in Table 2. For all HCCs, although the median washout time was ≥60 seconds for both VA and QA, the Wilcoxon signed-rank test yielded a standardized Z of −6.022 (P<0.001), indicating that washout was identified significantly earlier by QA than by VA. This trend remained consistent in both moderately and poorly differentiated HCC subgroups. For welldifferentiated HCCs, a direct comparison between VA and QA was not feasible. In our study, 51.5% (120/233) of FLLs showed washout within 60 seconds when assessed by QA, compared with 39.9% (93/233) by VA. The washout times of non-HCC malignancy and benign lesions assessed by QA and VA are presented in Table S1.

Distribution of CEUS LI-RADS categories based on VA and QA

The intraobserver and interobserver reliability of visual VA is summarized in Table S2. Interobserver agreement between the two radiologists for the CEUS LI-RADS classifications of all FLLs demonstrated good, with a weighted κ value of 0.80. Furthermore, the CEUS LI-RADS classifications by VA and QA also exhibited substantial agreement, with a weighted κ value of 0.79. The distribution of CEUS LI-RADS categories for FLLs based on VA and QA is presented in Table 3. The CEUS LI-RADS classifications based on VA and QA demonstrated almost perfect agreement for LR-1 (κ=1.00), LR-3 (κ=1.00), and LR-4 (κ=1.00) categories. However, VA and QA demonstrated moderate agreement for LR-5 and LR-M categories, with κ values of 0.74 and 0.73. Ninety-six lesions were classified as LR-5 by both VA and QA, and 91 lesions were classified as LR-M by both methods. However, 29 lesions were categorized as LR-5 by VA but classified as LR-M by QA (Figure S2), while 2 lesions were classified as LR-M by VA and as LR-5 by QA.

Diagnostic performance of QA LR-5 with different early washout threshold for HCC

The diagnostic performance of QA LR-5 across varying early washout thresholds for HCC is presented in Table 4 and Figure 5. When the early washout threshold was adjusted to 40 seconds, QA LR-5 achieved the best diagnostic performance, yielding an AUC of 0.818 [95% confidence interval (CI): 0.762–0.865] and a Youden’s index of 0.636. Consequently, 40 seconds was identified as the optimal early washout threshold for QA in HCC diagnosis. The modified QA LR-5, adopting this 40-second threshold, demonstrated a diagnostic accuracy of 80.7%, sensitivity of 79.9%, and specificity of 83.7%.

Figure 5 Diagnostic performance of QA for HCC using different early washout thresholds. Sensitivity (A), specificity (B), accuracy (C), AUC (D), and Youden’s index (E) in the diagnosis of HCC. A 40 seconds washout threshold yielded the highest AUC and Youden’s index for HCC diagnosis (red arrow). AUC, area under the curve; HCC, hepatocellular carcinoma; QA, quantitative analysis.

Diagnostic performance of VA LR-5, QA LR-5, and modified QA LR-5 for HCC

As mentioned above, a 40 seconds threshold was adopted as the modified early washout cut-off in this study, referred to as “modified QA LR-5”. The diagnostic performance of VA LR-5, QA LR-5, and modified QA LR-5 for HCC is exhibited in Table 5. QA LR-5 showed inferior sensitivity (51.6% vs. 66.3%, P<0.001), accuracy (60.5% vs. 72.1%, P<0.001) and AUC (0.728 vs. 0.801, P<0.001) compared with VA LR-5, though demonstrated comparable specificity (93.9% vs. 93.9%, P>0.05). However, compared with that of VA LR-5, the modified QA LR-5 exhibited superior sensitivity (79.9% vs. 66.3%, P<0.001), accuracy (80.7% vs. 72.1%, P<0.001) and AUC (0.818 vs. 0.801, P>0.05), while maintaining comparable specificity (83.7% vs. 93.9%, P>0.05).

Discussion

This study analyzed 233 FLLs from 225 patients at high risk for HCC, with both VA and QA used to assess washout time. Although the median washout time of FLLs occurred earlier with QA than with VA, the CEUS LI-RADS classifications exhibited good agreement between the two methods (weighted κ=0.79). Moreover, modifying the washout time threshold from 60 to 40 seconds improved the diagnostic sensitivity and accuracy of QA LR-5 for HCC.

Although VA offers a straightforward method for evaluating CEUS characteristics of FLLs, it is hampered by several limitations that QA can address. One of the primary limitations is its subjectivity. The operator’s focus may be drawn to the specific enhancing area rather than on a systematic comparison of the entire lesion’s echogenicity to the surrounding liver parenchyma (26,27). This subjectivity is particularly problematic in evaluating washout, as its appearance in some HCCs can be ambiguous, and the detection of mild washout is highly dependent on observer experience (28). Consequently, interobserver agreement for assessing washout in the portal or late phase is notably lower than for assessing arterial phase enhancement (28). Therefore, QA serves as a valuable objective tool to compensate for these inherent limitations of VA, particularly in the standardized assessment of washout. Growing clinical interest has been directed toward QA and its application in current and future practice. The 2023 EFSUMB Technical Review has highlighted its pivotal role, positing that advances in big data analytics and machine learning leveraging dynamic CEUS-QA could significantly strengthen the clinical value of CEUS (19). Notably, the 2024 US LI-RADS Surveillance update incorporated visualization scores into management recommendations, highlighting the value of standardized exam-quality assessment. Likewise, future AI-driven CEUS QA may enable CEUS-specific “visualization scores” to reduce operator dependence and improve patient stratification and management (29).

Although the median washout times for the 184 HCCs were comparable between QA and VA, their distributions exhibited a significant positional shift: in most paired comparisons, washout was detected earlier by QA than by VA. This difference was statistically significant (Z=−6.022, P<0.001). Subgroup analysis stratified by the degree of HCC differentiation grade showed that the moderately and poorly differentiated subgroups yielded conclusions consistent with those observed in the overall HCC cohort. However, for well-differentiated HCCs, all lesions exhibited washout times exceeding 60 seconds on both VA and QA. Given that QA only captured washout occurring within the first 60 seconds, a direct comparison between VA and QA in this subgroup was not feasible. Nevertheless, we hypothesize that, for well-differentiated HCCs, washout would also be identified earlier with QA than with VA, consistent with the trend observed in other subgroups. Furthermore, given that well differentiated HCCs typically exhibit delayed washout, whereas poorly differentiated HCCs tend to show earlier washout, this imbalance may potentially bias the overall washout kinetics toward an earlier onset. However, it is important to note that the primary objective of our study was to compare the diagnostic performance of VA versus QA in determining CEUS LI-RADS classification, using a paired design within the same lesion. In this paired comparative framework, the potential confounding effect of tumor differentiation is inherently minimized, as each lesion serves as its own control. Notably, we observed an interesting finding that, compared with moderately differentiated tumors, the discrepancy in washout time between QA and VA was reduced in poorly differentiated HCC. This pattern may be related to the underlying tumor biology. Earlier washout in biologically aggressive HCC may result from combined structural and hemodynamic alterations: as histologic differentiation decreases, the reduction in wide tumor sinusoids may allow faster intratumoral blood flow, while concomitant microvascular invasion may further disturb venous outflow and tumor microcirculation, thereby accelerating contrast clearance and contributing to earlier washout on CEUS (12,30).

Notably, with the rising prevalence of metabolic disorders, pseudo-washout in steatotic liver has become an increasingly relevant concern. Rather than reflecting true contrast clearance from the lesion, it likely results from prolonged enhancement and altered microbubble kinetics of the fatty background parenchyma, together with increased background harmonic signal (31). Therefore, in future investigations, hepatic steatosis should be prospectively documented, analyses should be stratified accordingly, and quantitative assessment of background parenchymal enhancement should be incorporated to better distinguish true washout from relative pseudo-washout.

Discordant CEUS LI-RADS classification was observed in 31 lesions, with a significant predominance of cases categorized as LR-5 by VA but LR-M by QA (n=29) over the reverse pattern (LR-M by VA but LR-5 by QA; n=2). Pathological confirmation of the 29 lesions in the former group revealed 28 HCCs and one intrahepatic cholangiocarcinoma, indicating that QA misclassified 28 HCCs as QA LR-M. This misclassification led to the reduced diagnostic accuracy of QA for HCC compared to VA. The discrepancy is attributable to the earlier washout detected by QA. The two lesions (categorized as LR-M by VA and LR-5 by QA) were pathologically diagnosed as HCC and carcinosarcoma respectively. They were classified as LR-5 by QA because they exhibited only a small portion of early washout, with the majority of the lesion remaining the significantly hyperenhanced. The washout present in a small portion of these two lesions can make it more noticeable visually due to the difference in brightness. This finding suggests that for lesions with partial washout, the applicability of QA still requires validation using a larger dataset.

The primary objective of CEUS LI-RADS is to achieve high specificity for HCC diagnosis in LR-5 (32), though this emphasis on specificity may come at the cost of sensitivity and accuracy. Since the existing CEUS LI-RADS was primarily designed for VA, applying its diagnostic criteria directly to QA may compromise the diagnostic performance of the original system. In this study, 40 seconds washout threshold was identified as the optimal cut-off for QA-based HCC diagnosis. Compared with VA LR-5, the modified QA LR-5 achieved significantly higher sensitivity and overall accuracy for diagnosing HCC, and a substantial proportion of HCCs initially classified as non-LR-5 by VA were correctly reclassified as QA LR-5. Importantly, this shift from the conventional 60-second threshold to 40 seconds likely reflects a fundamental difference between visual assessment and QA. In the present study, 40 seconds appeared to be a more suitable early washout threshold for injection-based QA; however, this finding requires further validation in larger multicenter cohorts. Moreover, because arterial contrast arrival may vary with individual hemodynamic status, future studies should also investigate arrival-adjusted temporal metrics, such as the AWTI, as potential refinements to quantitative CEUS assessment.

A considerable proportion of HCCs may show washout beyond 60 seconds, and some may even exhibit washout only in the late phase. Under the current CEUS LI-RADS criteria, 60 seconds serves as the threshold for distinguishing early from late washout. Nevertheless, prior studies have demonstrated that a subset of HCCs, particularly poorly differentiated lesions, can present with washout within 60 seconds (11,13,33). Consequently, these HCCs may be assigned to the CEUS LR-M category rather than LR-5, thereby reducing the sensitivity of CEUS LI-RADS for HCC diagnosis. On this basis, we used 60 seconds as the upper boundary of analysis to explore candidate thresholds before this time point and to determine whether an alternative cutoff could improve the diagnostic sensitivity of the CEUS LI-RADS framework. It is worth noting that the earlier washout time measured by QA may partly reflect its greater temporal sensitivity, as software-based analysis can detect subtle contrast intensity changes and identify washout onset more precisely than visual assessment. This methodological difference may help explain why a shorter washout threshold performed better in QA than in conventional visual assessment. In the present study, 40 seconds appeared to be a more suitable early washout threshold for injection-based QA; however, this finding requires further validation in larger multicenter cohorts. Moreover, because arterial contrast arrival may vary with individual hemodynamic status, future studies should also investigate arrival-adjusted temporal metrics, such as the arterial phase onset-to-washout interval, as potential refinements to quantitative CEUS assessment.

Several limitations should be acknowledged in this study. First, the sample size was relatively small because all enrolled cases were pathologically confirmed, however, this requirement also strengthened the study’s validity. Second, as continuous scanning during CEUS may cause excessive microbubble destruction, only dynamic images within the first 60 seconds were analyzed, although washout time cut-offs within this interval proved sufficient for QA to distinguish HCC from other non-HCC malignancies. Third, the number of well-differentiated HCCs in our cohort was relatively limited, which may have contributed to the earlier washout onset observed in this study. Future studies incorporating larger patient cohorts will be necessary to minimize potential bias and further validate our findings. Finally, the VueBox^® software used for QA offers motion compensation limited to in-plane respiratory or probe motion, but does not address more complex or out-of-plane movements (34).

Conclusions

The washout time of FLLs is earlier with QA than with VA, although both methods demonstrate good agreement in CEUS LI-RADS classifications. Modifying the washout threshold from 60 to 40 seconds significantly improves the diagnostic performance of QA. This modification is a critical step toward adapting the visually-established CEUS LI-RADS washout criterion for future AI-driven QA applications.

Acknowledgments

None.

Footnote

Reporting Checklist: The authors have completed the STARD reporting checklist. Available at https://qims.amegroups.com/article/view/10.21037/qims-2026-1-0144/rc

Data Sharing Statement: Available at https://qims.amegroups.com/article/view/10.21037/qims-2026-1-0144/dss

Funding: This work was supported by the National Natural Science Foundation of China (No. 82171952), the Youth Program of the National Natural Science Foundation of China (No. 82503212), as well as the Sichuan Natural Science Foundation (No. 2024NSFSC0043).

Conflicts of Interest: All authors have completed the ICMJE uniform disclosure form (available at https://qims.amegroups.com/article/view/10.21037/qims-2026-1-0144/coif). The authors have no conflicts of interest to declare.

Ethical Statement: The authors are accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved. This study was conducted in accordance with the Declaration of Helsinki and its subsequent amendments. The study was approved by the Ethics Committee of West China Hospital, Sichuan University (No. 2020-708) and the requirement for individual consent was waived due to the retrospective nature of the study.

Open Access Statement: This is an Open Access article distributed in accordance with the Creative Commons Attribution-NonCommercial-NoDerivs 4.0 International License (CC BY-NC-ND 4.0), which permits the non-commercial replication and distribution of the article with the strict proviso that no changes or edits are made and the original work is properly cited (including links to both the formal publication through the relevant DOI and the license). See: https://creativecommons.org/licenses/by-nc-nd/4.0/.

References

1. Bray F, Laversanne M, Sung H, Ferlay J, Siegel RL, Soerjomataram I, Jemal A. Global cancer statistics 2022: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin 2024;74:229-63. [Crossref] [PubMed]
2. Liu Y, Wei X, Zhang X, Pang C, Xia M, Du Y. CT radiomics combined with clinical variables for predicting the overall survival of hepatocellular carcinoma patients after hepatectomy. Transl Oncol 2022;26:101536. [Crossref] [PubMed]
3. Jo PC, Jang HJ, Burns PN, Burak KW, Kim TK, Wilson SR. Integration of Contrast-enhanced US into a Multimodality Approach to Imaging of Nodules in a Cirrhotic Liver: How I Do It. Radiology 2017;282:317-31. [Crossref] [PubMed]
4. Singal AG, Llovet JM, Yarchoan M, Mehta N, Heimbach JK, Dawson LA, Jou JH, Kulik LM, Agopian VG, Marrero JA, Mendiratta-Lala M, Brown DB, Rilling WS, Goyal L, Wei AC, Taddei TH. AASLD Practice Guidance on prevention, diagnosis, and treatment of hepatocellular carcinoma. Hepatology 2023;78:1922-65. [Crossref] [PubMed]
5. Atri M, Jang HJ, Kim TK, Khalili K. Contrast-enhanced US of the Liver and Kidney: A Problem-solving Modality. Radiology 2022;303:11-25. [Crossref] [PubMed]
6. Son JH, Choi SH, Kim SY, Lee SJ, Park SH, Kim KW, Won HJ, Shin YM, Kim PN. Accuracy of contrast-enhanced ultrasound liver imaging reporting and data system: a systematic review and meta-analysis. Hepatol Int 2020;14:1104-13. [Crossref] [PubMed]
7. Shin J, Lee S, Bae H, Chung YE, Choi JY, Huh YM, Park MS. Contrast-enhanced ultrasound liver imaging reporting and data system for diagnosing hepatocellular carcinoma: A meta-analysis. Liver Int 2020;40:2345-52. [Crossref] [PubMed]
8. Li J, Yang L, Ma L, Lu Q, Luo Y. Diagnostic Accuracy of Contrast-Enhanced Ultrasound Liver Imaging Reporting and Data System (CEUS LI-RADS) for Differentiating Between Hepatocellular Carcinoma and Other Hepatic Malignancies in High-Risk Patients: A Meta-Analysis. Ultraschall Med 2021;42:187-93. [Crossref] [PubMed]
9. Kim TK, Noh SY, Wilson SR, Kono Y, Piscaglia F, Jang HJ, Lyshchik A, Dietrich CF, Willmann JK, Vezeridis A, Sirlin CB. Contrast-enhanced ultrasound (CEUS) liver imaging reporting and data system (LI-RADS) 2017 - a review of important differences compared to the CT/MRI system. Clin Mol Hepatol 2017;23:280-9. [Crossref] [PubMed]
10. Anaye A, Perrenoud G, Rognin N, Arditi M, Mercier L, Frinking P, Ruffieux C, Peetrons P, Meuli R, Meuwly JY. Differentiation of focal liver lesions: usefulness of parametric imaging with contrast-enhanced US. Radiology 2011;261:300-10. [Crossref] [PubMed]
11. Qiu S, Ding J, Qin Z, Zhou Y, Zhou H, Wang Y, Zhao L, Jing X. Application of DCE-US using the LI-RADS for patients with liver nodules at high risk for hepatocellular carcinoma: A preliminary study and comparison with visual interpretation. Eur J Radiol 2022;155:110473. [Crossref] [PubMed]
12. Yang HK, Burns PN, Jang HJ, Kono Y, Khalili K, Wilson SR, Kim TK. Contrast-enhanced ultrasound approach to the diagnosis of focal liver lesions: the importance of washout. Ultrasonography 2019;38:289-301. [Crossref] [PubMed]
13. Zheng W, Li Q, Zou XB, Wang JW, Han F, Li F, Huang LS, Li AH, Zhou JH. Evaluation of Contrast-enhanced US LI-RADS version 2017: Application on 2020 Liver Nodules in Patients with Hepatitis B Infection. Radiology 2020;294:299-307. [Crossref] [PubMed]
14. Huang JY, Li JW, Lu Q, Luo Y, Lin L, Shi YJ, Li T, Liu JB, Lyshchik A. Diagnostic Accuracy of CEUS LI-RADS for the Characterization of Liver Nodules 20 mm or Smaller in Patients at Risk for Hepatocellular Carcinoma. Radiology 2020;294:329-39. [Crossref] [PubMed]
15. Ding J, Qin Z, Zhou Y, Zhou H, Zhang Q, Wang Y, Jing X, Wang F. Impact of Revision of the LR-M Criteria on the Diagnostic Performance of Contrast-Enhanced Ultrasound LI-RADS. Ultrasound Med Biol 2021;47:3403-10. [Crossref] [PubMed]
16. Wang Y, Zhu Z, Zhu M, Wu S. Arterial-Washout Temporal Profiling in CEUS LI-RADS: A Diagnostic Algorithm for Reducing Hepatocellular Carcinoma Misclassification. Radiol Imaging Cancer 2026;8:e250259. [Crossref] [PubMed]
17. Schellhaas B, Hammon M, Strobel D, Pfeifer L, Kielisch C, Goertz RS, Cavallaro A, Janka R, Neurath MF, Uder M, Seuss H. Interobserver and intermodality agreement of standardized algorithms for non-invasive diagnosis of hepatocellular carcinoma in high-risk patients: CEUS-LI-RADS versus MRI-LI-RADS. Eur Radiol 2018;28:4254-64. [Crossref] [PubMed]
18. Zhou H, Zhang C, Du L, Jiang J, Zhao Q, Sun J, Li Q, Wan M, Wang X, Hou X, Wen Q, Liu Y, Zhou X, Huang P. Contrast-Enhanced Ultrasound Liver Imaging Reporting and Data System in Diagnosing Hepatocellular Carcinoma: Diagnostic Performance and Interobserver Agreement. Ultraschall Med 2022;43:64-71. [Crossref] [PubMed]
19. Dietrich CF, Correas JM, Cui XW, Dong Y, Havre RF, Jenssen C, Jung EM, Krix M, Lim A, Lassau N, Piscaglia F. EFSUMB Technical Review - Update 2023: Dynamic Contrast-Enhanced Ultrasound (DCE-CEUS) for the Quantification of Tumor Perfusion. Ultraschall Med 2024;45:36-46. [Crossref] [PubMed]
20. Wiesinger I, Jung F, Jung EM. Contrast-enhanced ultrasound (CEUS) and perfusion imaging using VueBox®. Clin Hemorheol Microcirc 2021;78:29-40. [Crossref] [PubMed]
21. Pei XQ, Liu LZ, Xiong YH, Zou RH, Chen MS, Li AH, Cai MY. Quantitative analysis of contrast-enhanced ultrasonography: differentiating focal nodular hyperplasia from hepatocellular carcinoma. Br J Radiol 2013;86:20120536. [Crossref] [PubMed]
22. Kono Y, Lyshchik A, Cosgrove D, Dietrich CF, Jang HJ, Kim TK, Piscaglia F, Willmann JK, Wilson SR, Santillan C, Kambadakone A, Mitchell D, Vezeridis A, Sirlin CB. Contrast Enhanced Ultrasound (CEUS) Liver Imaging Reporting and Data System (LI-RADS®): the official version by the American College of Radiology (ACR). Ultraschall Med 2017;38:85-6. [Crossref] [PubMed]
23. Kratzer W, Güthle M, Dobler F, Seufferlein T, Graeter T, Schmidberger J, Barth TF, Klaus J. Comparison of superb microvascular imaging (SMI) quantified with ImageJ to quantified contrast-enhanced ultrasound (qCEUS) in liver metastases-a pilot study. Quant Imaging Med Surg 2022;12:1762-74. [Crossref] [PubMed]
24. Shin J, Lee S, Kim YY, Chung YE, Choi JY, Park MS. Contrast-enhanced ultrasound Liver Imaging Reporting and Data System category M: a systematic review and meta-analysis. Ultrasonography 2022;41:74-82. [Crossref] [PubMed]
25. Dong Y, Qiu Y, Yang D, Yu L, Zuo D, Zhang Q, Tian X, Wang WP, Jung EM. Potential application of dynamic contrast enhanced ultrasound in predicting microvascular invasion of hepatocellular carcinoma. Clin Hemorheol Microcirc 2021;77:461-9. [Crossref] [PubMed]
26. Quaia E. State of the Art: LI-RADS for Contrast-enhanced US. Radiology 2019;293:4-14. [Crossref] [PubMed]
27. Schwarz S, Clevert DA, Ingrisch M, Geyer T, Schwarze V, Rübenthaler J, Armbruster M. Quantitative Analysis of the Time-Intensity Curve of Contrast-Enhanced Ultrasound of the Liver: Differentiation of Benign and Malignant Liver Lesions. Diagnostics (Basel) 2021;11:1244. [Crossref] [PubMed]
28. Quaia E, Alaimo V, Baratella E, Pizzolato R, Cester G, Medeot A, Cova MA. Effect of observer experience in the differentiation between benign and malignant liver tumors after ultrasound contrast agent injection. J Ultrasound Med 2010;29:25-36. [Crossref] [PubMed]
29. Kamaya A, Fetzer DT, Seow JH, Burrowes DP, Choi HH, Dawkins AA, Fung C, Gabriel H, Hong CW, Khurana A, McGillen KL, Morgan TA, Sirlin CB, Tse JR, Rodgers SK. LI-RADS US Surveillance Version 2024 for Surveillance of Hepatocellular Carcinoma: An Update to the American College of Radiology US LI-RADS. Radiology 2024;313:e240169. [Crossref] [PubMed]
30. Zhu L, Li N, Liang S, Jiang B, Li M, Luo Y, Fei X. High frame rate contrast-enhanced ultrasound in hepatocellular carcinoma: biological characteristics and patient outcomes. BMC Cancer 2025;25:1488. [Crossref] [PubMed]
31. Bertin C, Egels S, Wagner M, Huynh-Charlier I, Vilgrain V, Lucidarme O. Contrast-enhanced ultrasound of focal nodular hyperplasia: a matter of size. Eur Radiol 2014;24:2561-71. [Crossref] [PubMed]
32. Hu YX, Shen JX, Han J, Mao SY, Mao RS, Li Q, Li F, Guo ZX, Zhou JH. Diagnosis of Non-Hepatocellular Carcinoma Malignancies in Patients With Risks for Hepatocellular Carcinoma: CEUS LI-RADS Versus CT/MRI LI-RADS. Front Oncol 2021;11:641195. [Crossref] [PubMed]
33. Li F, Li Q, Liu Y, Han J, Zheng W, Huang Y, Zheng X, Cao L, Zhou JH. Distinguishing intrahepatic cholangiocarcinoma from hepatocellular carcinoma in patients with and without risks: the evaluation of the LR-M criteria of contrast-enhanced ultrasound liver imaging reporting and data system version 2017. Eur Radiol 2020;30:461-70. [Crossref] [PubMed]
34. Tranquart F, Mercier L, Frinking P, Gaud E, Arditi M. Perfusion quantification in contrast-enhanced ultrasound (CEUS)--ready for research projects and routine clinical use. Ultraschall Med 2012;33:S31-8. [Crossref] [PubMed]