Low–mechanical-index B-mode phase-inversion harmonic imaging significantly improves hepatic lesion conspicuity during Sonazoid Kupffer phase: a prospective comparative study

Article information

Ultrasonography. 2026;45(2):163-173

Publication date (electronic) : 2026 January 5

doi : https://doi.org/10.14366/usg.25176

Jeongin Yoo¹, Jeong Min Lee^,1,2, Hyo-Jin Kang¹, Jae Seok Bae¹

¹Department of Radiology, Seoul National University Hospital, Seoul National University College of Medicine, Seoul, Korea

²Institute of Radiation Medicine, Seoul National University Medical Research Center, Seoul, Korea

Correspondence to: Jeong Min Lee, MD, Department of Radiology and Institute of Radiation Medicine, Seoul National University College of Medicine, 101Daehak-ro, Jongno-gu, Seoul 03080, Korea Tel. +82-2-2072-3154 Fax. +82-2-743-6385 E-mail: jmlshy2000@gmail.com

Received 2025 September 3; Revised 2025 December 17; Accepted 2026 January 5.

Abstract

Graphical abstract

Introduction

Accurate detection and characterization of focal liver lesions, particularly hepatocellular carcinoma (HCC) and metastatic disease, remain critical challenges in diagnostic imaging and interventional radiology. The clinical demand for precise lesion targeting using ultrasonography has intensified with the expanding role of image-guided interventions in the management of hepatic malignancies. Procedures such as percutaneous biopsy, radiofrequency ablation, and microwave ablation require accurate real-time visualization not only of the target lesion but also of interventional devices and surrounding anatomical structures. Accordingly, improving lesion conspicuity is essential for early tumor detection and for safely guiding interventional procedures, particularly in patients with complex backgrounds such as cirrhosis.

Contrast-enhanced ultrasonography (CEUS) has advanced lesion characterization by providing dynamic vascular information using microbubble contrast agents [1,2]. Among currently available contrast agents, Sonazoid (perfluorobutane, GE Healthcare, Oslo, Norway) offers a distinct advantage through its Kupffer-phase (KP) imaging capability, which is enabled by selective uptake into hepatic Kupffer cells [3]. During this delayed postvascular phase (approximately 10–60 minutes after injection), malignant lesions that lack Kupffer cells appear as hypoenhancing defects against the enhanced liver parenchyma [2,4–8]. In clinical practice, KP imaging with Sonazoid-enhanced ultrasonography can depict smaller tumors than conventional B-mode ultrasonography; prior studies have shown that KP imaging detects lesions that are inconspicuous on standard ultrasonography, thereby improving early HCC detection rates [5]. Furthermore, because of the prolonged imaging window provided by the KP, Sonazoid-enhanced ultrasonography is often preferred over SonoVue-enhanced ultrasonography for guiding interventional procedures such as biopsy or ablation [1,4,8]. Consequently, Sonazoid-enhanced ultrasonography is now recommended in the Japanese Society of Hepatology guidelines as a second-line or even first-line imaging tool for HCC surveillance in patients with cirrhosis [9,10].

Despite these advantages, conventional CEUS during the KP—most commonly performed using amplitude modulation or standard phase-inversion (PI) techniques at a low mechanical index (MI)—has notable limitations [11]. Although conventional CEUS provides strong contrast enhancement, inherent constraints related to spatial resolution and suppression of tissue harmonic components can limit its utility for interventional applications. These limitations may result in reduced anatomical detail and suboptimal visualization of biopsy needles or ablation electrodes. To address these challenges, this study implemented a low-MI PI harmonic imaging (low-MI PIHI) technique that integrates both tissue and microbubble harmonic signals, thereby enhancing lesion conspicuity while preserving parenchymal texture [12]. It was hypothesized that the low-MI PIHI technique, which preserves tissue harmonics across a broader frequency bandwidth and thereby enhances border definition, maintains parenchymal texture, reduces noise, and improves lateral resolution, would demonstrate superior lesion conspicuity and margin delineation compared with conventional CEUS harmonic techniques during the KP [13]. Therefore, this study aimed to compare the imaging performance of low-MI PIHI with that of conventional CEUS harmonic imaging during the KP of Sonazoid-enhanced ultrasonography for visualization of hepatic lesions, focusing on lesion conspicuity and overall image quality.

Materials and Methods

Study Population

Between February 2023 and January 2024, 84 patients were initially enrolled after meeting the following inclusion criteria: (1) age ≥18 years; (2) high-risk status defined by the presence of cirrhosis or chronic hepatitis B; (3) at least one noncystic focal liver observation measuring ≥1 cm and suspicious for HCC; and (4) referral to the radiology department for CEUS to characterize focal liver observations, guide biopsy, or evaluate the feasibility of ablation (Fig. 1). The exclusion criteria were (1) a suboptimal image acquisition protocol during the adjustment period (n=10) and (2) poor delineation of suspicious observations on baseline B-mode ultrasonography (n=3). Ultimately, a total of 71 patients with 72 focal liver observations constituted the final study population. Patient demographics and baseline lesion characteristics are summarized in Table 1.

Fig. 1.

Flow chart of patient enrollment.

The diagram illustrates the application of inclusion and exclusion criteria used to derive the final study population from an initial cohort of 84 patients. HCC, hepatocellular carcinoma; CEUS, contrast-enhanced ultrasonography.

Table 1.

Patient demographics and baseline lesion characteristics

Sample size determination

A priori sample size calculation was performed using G*Power software (version 3.1) [14]. Based on a pilot study of 10 patients demonstrating a mean difference of 1.0 point in conspicuity scores (standard deviation, 1.2) between conventional KP imaging and low-MI PIHI KP imaging, it was estimated that 64 lesions would be required to detect a clinically meaningful difference of 0.5 points in conspicuity scores with 80% statistical power and a two-sided α of 0.05 using the Wilcoxon signed-rank test. To account for potential technical failures or screening exclusions, 71 patients were ultimately enrolled.

Imaging Protocol

All patients underwent CEUS examinations using Sonazoid, which was prepared according to the manufacturer’s instructions and administered intravenously at a dose of 0.015 mL/kg, followed by a 10-mL normal saline flush. Ultrasound examinations were performed using a commercially available ultrasound system (Aplio i800, Canon Medical Systems, Tochigi, Japan) equipped with a convex array transducer (i8CX1). All CEUS scans were conducted by one of two board-certified radiologists, each with more than 5 years of experience in hepatic CEUS. For image acquisition, baseline B-mode ultrasonography was initially performed to identify and document the target lesion. Continuous video recording was then obtained during the first minute after Sonazoid injection to capture the arterial phase, followed by single-frame image acquisitions at 2 minutes for portal venous phase and at 3 and 5 minutes for late vascular phase [1]. At 10 minutes after contrast injection, corresponding to the KP, two imaging modes were applied sequentially: conventional contrast-specific harmonic imaging (standard-MI CEUS mode) followed immediately by low-MI PIHI (Fig. 2).

Fig. 2.

Representative example of the contrast-enhanced ultrasonography study protocol in a 73-year-old woman with a 2.1-cm hepatocellular carcinoma located in liver segment III.

A. Baseline B-mode ultrasonography demonstrates a poorly delineated lesion in segment III (arrows). B. Arterial-phase image acquired with continuous imaging demonstrates nonrim hyperenhancement (arrows). C. Portal venous–phase image obtained 2 minutes after contrast injection shows late mild washout (arrows). D. Late vascular–phase image at 5 minutes demonstrates persistent washout (arrows). E. Kupffer-phase image at 10 minutes acquired using conventional contrast-enhanced imaging shows the lesion as hypoechoic (arrows). F. Kupffer-phase image at 10 minutes acquired using low–mechanical-index phase-inversion harmonic imaging demonstrates superior lesion conspicuity and sharper border delineation (arrows). This case illustrates the complete imaging protocol across all vascular phases and highlights the comparison between conventional and low–mechanical-index imaging techniques during the Kupffer phase.

Conventional CEUS harmonic imaging was performed using an MI of 0.2–0.25 with a dedicated contrast-specific pulse sequence. In contrast, the low-MI PIHI technique employed a modified B-mode configuration based on a differential tissue harmonic pulse sequence with a transmit frequency of 4 MHz. The low-MI PIHI parameters were optimized with an MI of 0.2, gain of 85%, dynamic range of 60 dB, frame rate of 24 frames per second, and spatial compounding (ApliPure+) set to 2. Other imaging settings, including real-time gain optimization (Quick Scan) and speckle reduction, were kept similar to the standard abdominal B-mode preset to preserve anatomical detail. This approach differs from conventional CEUS modes, which actively suppress background tissue signals.

Results

Lesion Conspicuity and Overall Image Quality Comparison

The study demonstrated a clear, progressive improvement in lesion conspicuity across the three imaging techniques (Table 2). For reviewer 1, median lesion conspicuity scores increased sequentially from 3.0 (interquartile range [IQR], 2.0 to 4.0) with baseline B-mode imaging to 4.0 (IQR, 3.0 to 4.0) with conventional KP imaging, and further to 5.0 (IQR, 4.0 to 5.0) with low-MI PIHI KP imaging (all pairwise comparisons, P<0.001). A similar progressive trend was observed for reviewer 2, with median scores increasing from 3.0 (IQR, 2.0 to 3.5) with B-mode imaging to 2.0 (IQR, 2.0 to 3.5) with conventional KP imaging, followed by a marked increase to 4.0 (IQR, 4.0 to 5.0) with low-MI PIHI KP imaging (all pairwise comparisons, P<0.001) (Figs. 2, 3).

Table 2.

Comparison of image quality, margin delineation, and lesion conspicuity between B-mode and low-MI PIHI KP imaging and between conventional and low-MI PIHI KP imaging

Table 3.

Comparison of lesion conspicuity between conventional CEUS imaging and low-MI PIHI during the Kupffer phase

Fig. 3.

Computed tomography and contrast-enhanced ultrasonography images of a 66-year-old man with a 1.5-cm hepatocellular carcinoma in liver segment VI.

A, B. Contrast-enhanced computed tomography demonstrates the lesion (arrows) with nonrim hyperenhancement in the arterial phase (A) and washout in the portal venous phase (B). C. Baseline B-mode ultrasonography shows poor lesion conspicuity (arrows; conspicuity score, 1–2). D. Conventional contrast-enhanced imaging at 10 minutes (Kupffer phase) depicts the lesion as a hypoechoic defect with moderate conspicuity (arrows). E. Low–mechanical-index phase-inversion harmonic imaging at 10 minutes (Kupffer phase) demonstrates significantly improved lesion conspicuity with sharper border delineation and enhanced lesion-to-liver contrast (arrows).

Lesion border delineation demonstrated a similar pattern of progressive improvement across imaging techniques. For reviewer 1, median scores increased from 3.0 (IQR, 2.0 to 3.0) with B-mode imaging to 3.0 (IQR, 3.0 to 4.0) with conventional KP imaging, and further to 4.0 (IQR, 4.0 to 5.0) with low-MI PIHI KP imaging. Reviewer 2 likewise showed improvement, with scores increasing from 3.0 (IQR, 2.0 to 3.0) with B-mode imaging to 2.0 (IQR, 2.0 to 3.0) with conventional KP imaging, and then to 4.0 (IQR, 3.0 to 5.0) with low-MI PIHI KP imaging (all pairwise comparisons, P<0.001) (Fig. 3).

Overall image quality followed the same progressive pattern. Median scores for both reviewers were 3.0 with baseline B-mode imaging, increased to 3.0 with conventional KP imaging (IQR, 3.0 to 4.5 for reviewer 1 and 2.0 to 3.0 for reviewer 2), and further increased to 4.0 with low-MI PIHI KP imaging (IQR, 3.0 to 4.5 for reviewer 1 and 3.0 to 4.0 for reviewer 2). These differences were statistically significant for both reviewers (P=0.003 for reviewer 1 and P<0.001 for reviewer 2).

Interobserver Agreement

Interobserver agreement for lesion conspicuity was substantial for baseline B-mode imaging (κ=0.677; 95% confidence interval [CI], 0.568 to 0.787), moderate for low-MI PIHI KP imaging (κ=0.481; 95% CI, 0.220 to 0.743), and poor for conventional KP imaging (κ=0.175; 95% CI, 0.041 to 0.309) (Table 4).

Table 4.

Interobserver agreement of image quality, margin delineation, and lesion conspicuity between two reviewers on B-mode, conventional KP, and low-MI PIHI KP imaging

Discussion

This study demonstrates that the low-MI PIHI technique significantly improves hepatic lesion conspicuity and margin delineation compared with conventional harmonic CEUS during the Sonazoid KP. The marked improvement in lesion conspicuity, with median scores increasing from 4.0 and 2.0 to 5.0 and 4.0, respectively, for the two reviewers (P<0.001), may translate into enhanced diagnostic confidence for lesion localization and improved treatment precision. The technique also demonstrated superior performance across different lesion sizes, with particularly pronounced improvements for smaller lesions (≤2 cm), most evident in reviewer 2’s assessments (from 2.0 to 4.0, P<0.001). This improvement addresses a critical clinical need, as patients with cirrhosis represent the highest-risk population for HCC development and frequently present the greatest challenges for lesion detection and localization during interventional procedures.

The clinical advantages of enhanced lesion conspicuity achieved with low-MI PIHI KP imaging are multifold and directly relevant to patient care. Improved lesion visibility with preserved anatomical detail facilitates precise lesion localization and targeting during image-guided procedures and may assist in determining optimal access routes, reducing sampling error, decreasing the need for multiple needle passes, and potentially shortening procedure time [4,15]. Although conventional harmonic imaging during the KP of Sonazoid-enhanced ultrasonography is effective for maximizing contrast between lesions and surrounding liver parenchyma [5,15], it presents challenges for interventional guidance. The strong suppression of background tissue signals, which underlies its contrast enhancement, inherently compromises spatial resolution, obscures fine anatomical detail, and limits visualization of biopsy needles or ablation electrodes [11]. To overcome these limitations during procedures such as biopsy or ablation, operators often switch to conventional B-mode imaging at a high MI (>1.5), which restores anatomical detail necessary for instrument guidance. However, this approach entails a trade-off: high-MI ultrasonography pulses intentionally destroy contrast microbubbles retained within Kupffer cells, resulting in a characteristic artifact known as the “migrating echogenic band,” which appears as wave-like, inhomogeneous echo enhancement sweeping across the image [1]. In contrast, preserved anatomical detail and improved visualization of instruments, such as biopsy needles or ablation electrodes, on low-MI PIHI KP imaging enhance procedural precision and may also contribute to reducing complications related to needle or electrode placement, including bleeding or vascular injury [11]. Although these procedural benefits were not systematically evaluated in the present study, the authors’ subsequent clinical experience using this technique to guide biopsies and radiofrequency ablations has been favorable. The authors have anecdotally observed clearer visualization of needles and electrodes against the preserved parenchymal background, along with maintained lesion border delineation during procedures. These observations strongly support the need for a dedicated future study to formally quantify procedural advantages.

The superior performance of low-MI PIHI likely arises from fundamental differences in signal processing and frequency utilization compared with conventional CEUS techniques. Low-MI PIHI simultaneously captures and integrates microbubble-derived harmonic signals and tissue harmonic components across a broader effective frequency bandwidth [11]. This dual-harmonic integration preserves fine anatomical detail while maintaining contrast sensitivity, thereby mitigating the inherent trade-off between spatial resolution and contrast enhancement that characterizes conventional CEUS modes. Conventional contrast-specific harmonic imaging techniques prioritize isolation of microbubble signals through aggressive suppression of tissue harmonics, which necessarily compromises spatial resolution and anatomical detail. These approaches typically rely on amplitude modulation or standard PI using relatively narrow frequency windows optimized for microbubble resonance [11]. In contrast, low-MI PIHI operates under a different paradigm, leveraging complementary information from both tissue and microbubble signals. The technique employs PI with an expanded frequency bandwidth encompassing both microbubble resonance frequencies and tissue harmonic components generated through nonlinear acoustic propagation [16]. In practical terms, operating at a low MI within a B-mode–based framework allows the use of features such as spatial compounding and speckle-reduction algorithms, which are often restricted in dedicated CEUS presets. This configuration supports image sharpness and depiction of anatomical structures rather than solely enhancing microbubble contrast. Preservation of characteristic hepatic parenchymal speckle further aids lesion-liver discrimination, an advantage that is particularly relevant in cirrhotic livers where heterogeneous background echogenicity complicates lesion detection [11,13].

Implementation of low-MI PIHI for KP imaging is straightforward, requiring only adjustment of scanner settings to maintain a low MI and relatively high gain, without the need for additional hardware, thereby allowing seamless integration into existing Sonazoid KP workflows. Nevertheless, several technical considerations warrant attention. This technique is currently optimized for the KP of Sonazoid-enhanced ultrasonography, which limits its applicability to contrast agents lacking postvascular phase retention. Because tissue harmonic components are intentionally preserved, the pure contrast-to-tissue signal ratio may be lower than that achieved with fully tissue-suppressed modes, potentially complicating detection of lesions with minimal signal differences [3,11,13]. Appropriate parameter optimization, particularly with respect to MI and gain settings, is therefore essential to maintain adequate contrast resolution [12]. In addition, higher-frequency harmonic components are more susceptible to attenuation, which may reduce visualization of deeply located lesions or affect performance in patients with higher body mass index [4,17]. Another practical consideration is the generalizability of this approach across ultrasound platforms. Although this study was conducted using a Canon Aplio i800 system, the fundamental principle of applying a very low-MI, B-mode–based harmonic imaging setting during the KP—rather than a dedicated, tissue-suppressed CEUS mode—is not vendor-specific. Similar low-MI phase or pulse inversion harmonic imaging modes are available on other major ultrasound platforms, including those from GE Healthcare and Samsung Medison, and could potentially be optimized for this application. While specific preset names and optimization parameters will vary by manufacturer, the authors believe that the concept of “B-mode–like” KP imaging is broadly achievable, although vendor-specific validation studies are required.

An interesting finding from the subgroup analysis was the differential response observed between HCC and non-HCC lesions, although these results should be interpreted cautiously given the limited number of non-HCC lesions (n=14, representing only 19.4% of total lesions). Among non-HCC lesions, reviewer 1 showed no significant improvement, with a median score of 5.0 for both imaging techniques, whereas reviewer 2 demonstrated a significant improvement, resulting in notable inter-reader variability. The absence of improvement observed by reviewer 1 may reflect a ceiling effect, whereby non-HCC lesions were already maximally visible with conventional imaging due to their distinct pathophysiological characteristics. Specifically, cholangiocarcinomas and metastases, which comprised the majority of non-HCC lesions in the present cohort, characteristically lack Kupffer cells entirely, leading to marked hypoenhancement during the KP that renders these lesions readily apparent even with conventional imaging techniques [1,2,7,18]. This pronounced washout pattern may leave limited room for further enhancement of lesion conspicuity. However, the discordant findings between reviewers suggest that this ceiling effect may not be universal and could be influenced by individual observer experience and detection thresholds. Moreover, the small sample size of non-HCC lesions limits the statistical power of this subgroup analysis and may not fully capture the spectrum of non-HCC malignancies encountered in clinical practice, underscoring the need for larger studies to draw definitive conclusions regarding the efficacy of this technique in this population.

Several limitations of the present study warrant consideration. First, the investigation focused on diagnostic performance metrics, specifically lesion conspicuity and delineation, rather than direct procedural utility. This study did not conduct a systematic evaluation of practical performance during interventional procedures, nor did it assess objective clinical outcomes such as diagnostic accuracy, biopsy yield, or completeness of ablation margins, which are necessary to definitively establish clinical superiority and remain important areas for future research. Second, this study was performed at a single tertiary referral center with specialized expertise in liver imaging, which may limit the generalizability of the findings to community practice settings. Although the sample size of 71 patients was sufficient for technical comparison, it may not have encompassed the full spectrum of lesion characteristics encountered in routine clinical practice, particularly for less common lesion types, as only 14 non-HCC lesions were included. In addition, the absence of blinded independent readers and the lack of standardized quantitative metrics for lesion conspicuity assessment introduce the potential for observer bias. Future investigations should address these limitations through multicenter prospective validation studies involving more diverse patient populations, standardized conspicuity measurements, and clinically meaningful endpoints, including procedural success rates and treatment monitoring efficacy. Furthermore, comparative analyses across a broader range of lesion subtypes, sizes, and depths would help define optimal clinical application scenarios for this technique. Integration with advanced imaging approaches, such as radiomics or fusion imaging, may further enhance clinical utility and represents a promising direction for future research [19,20].

In conclusion, low-MI PIHI significantly enhances hepatic lesion conspicuity and margin delineation compared with conventional CEUS harmonic imaging during the Sonazoid KP, with concomitant improvements in overall image quality. These enhancements may support more confident lesion detection and facilitate image-guided procedures in clinical practice.

Article information Continued

Notes