High-frequency ultrasound for microcirculation in metastatic mouse lymph node: contrast-free versus contrast-enhanced imaging

Abstract

Focal defects are one of the important features for the diagnosis of lymph node (LN) metastasis. In our previous study, an accurate method for detecting contrast agents was proposed. However, conventional B-mode and contrast-enhanced images via ultrasound contrast agents (UCAs) have the limitations of contrast and spatial resolution to visualize the microcirculation, such as focal defects to distinguish between benign and malignant LN. In the present study, we have developed a novel method based on clutter filtering with singular value decomposition (SVD) analysis using time-integrated amplitude envelope (TIAE) in high-frequency 40 MHz ultrasound for high contrast resolution of the microcirculation in LN tissue. A mouse LN was visualized in vivo without and with UCA to compare the contrast enhancement. A metastatic LN model was established with LM8-luc cells of C3H/HeJ-lpr/lpr and MXH54/Mo-lpr/lpr mice. Bioluminescence imaging and pathological observations were also conducted to evaluate tumor growth. It was found that clutter-filtered contrast-enhanced images with UCA could visualize the feature of the microcirculation in the control LN and focal defects in the metastatic LN. Consistent with histological findings of disrupted architecture and cellular heterogeneity, whereas clutter-filtered B-mode images without UCA failed to visualize the vascular circulation. TIAE provided images with high noise resistance, and the calculated vascular area in the LN showed a decreasing trend in the metastatic group compared to the control group. Our framework enables robust visualization and quantification of LN heterogeneity in microcirculation.

Introduction

Accurate evaluation of sentinel lymph node (LN) status is essential for cancer staging, prognosis prediction and treatment planning in malignancies such as breast cancer, head and neck cancer, and cutaneous tumors^1,2. Recent studies have reported that metastatic LNs often serve as origins of distant metastases^3,4, further emphasizing the clinical significance of assessing LN characteristics. Hallmarks of early metastatic involvement such as subtle tissue alterations, localized cortical thinning and irregular margins require high-resolution and high-contrast visualization, which standard imaging modes often fail to provide.

Ultrasound imaging has become the modality of choice in clinical settings owing to its real-time capability, non-invasiveness and repeatability⁵. Additionally, its portability and cost-effectiveness allow for immediate use during surgery or at the bedside, facilitating widespread application in LN evaluation. However, conventional B-mode ultrasound imaging has limited capacity to detect micrometastases and early structural changes within LNs⁶. Focal defects are one of the important features in diagnosing LN metastasis. Contrast-enhanced ultrasound via ultrasound contrast agents (UCAs) is useful to visualize the microcirculation to confirm the focal defects. Our previous study has proposed an accurate method for detecting contrast agents⁷. However, conventional contrast-enhanced images have the limitations of contrast and spatial resolution to visualize the microcirculation such as focal defects to distinguish between benign and malignant LNs⁸.

A major factor degrading the contrast resolution is signal noise collectively termed clutter. Clutter arises from multiple physical phenomena in the soft tissue, including off-axis scattering, sidelobe interference, multiple reflections and probe-internal reverberations. These pseudo-echo components obscure true reflection signals, thereby reducing image contrast and impairing boundary visualization⁹. To address this issue, various signal processing algorithms, such as adaptive beamforming¹⁰, spatial coherence analysis¹¹, principal component analysis¹² and slow-time filtering^{13, 14–15} have been proposed.

Our previous study developed a radio-frequency (RF) data acquisition system for in vivo measurement using high-frequency ultrasound¹⁶. In this system, the feature of tumor tissue in the LN could be characterized as the change in the attenuation coefficient. In the present study, we applied a developed clutter reduction technique using In-phase/Quadrature-phase (IQ) data obtained from the collected RF data with high-frequency at 40 MHz and evaluated its effectiveness in mice LNs. This method enhances flowing components based on the difference in spatial and temporal frequencies, efficiently removing clutter while preserving both phase and amplitude information. This preservation enables highly precise imaging of the microcirculation. Furthermore, we applied this method to metastatic LNs in the mouse model and quantitatively assessed its diagnostic utility by extracting focal defects observed during early metastatic involvement and correlating them with histopathological findings^{17, 18–19}.

Results

As illustrated in Fig. 1, we established a metastatic model by inoculating tumor cells into the subiliac LN (SiLN), inducing metastasis in the proper axillary LN (PALN) via lymphatic drainage (Fig. 1(A)). High-frequency ultrasound RF signals were recorded both before and after systemic administration of UCA. Frame-by-frame differential analysis revealed pulsatile intensity changes localized within the cardiac cycle (Figs. 1(B-1), 1(B-2) and 1(B-3)), indicating dynamic signal fluctuations primarily associated with cardiac pulsation. No significant differences were noted between pre- and post-pulsation phases, underscoring the specificity of the detected motion to pulsatile activity.

Fig. 1 [Images not available. See PDF.]

Schematic view of experimental design and signal analysis. Tumor cells were inoculated into the SiLN to induce metastasis in the PALN. In ultrasound data acquisition, RF signals were acquired before and after UCA injection (A). The effect of cardiac motion was confirmed in the temporal difference of amplitude envelope (B). (B-1) shows B-mode image intensity; (B-2) depicts frame-by-frame intensity differences highlighting a cardiac cycle; (B-3) illustrates the difference between reference (5th frame) and specific frames in pre-, during-, and post-pulsation phases.

Second, we confirmed how contrast-free and contrast-enhanced imaging of high-frequency ultrasound with clutter filtering enhanced the visibility of vascular structure in non-metastatic and metastatic LNs. Figures 2A-C display the amplitude envelope at the 1st frame, time-integrated amplitude envelope (TIAE), and detected vascular areas by thresholding in non-metastatic LN without or with UCA. Singular value decomposition (SVD) effectively separated tissue and flow components such as UCA and vascular areas. Figure 3 also displays the case in metastatic LN. The overall vascular areas such as shown in Fig. 3(C-3) decreased around 8% compared to Fig. 2(C-3). In addition, the scattered vascular distribution was observed on the right side of the LN as well as low echogenicity of tissues.

Fig. 2 [Images not available. See PDF.]

Visualization of vascular structures in control LN. (A–C) show the B-mode without clutter filter, TIAE map, and overlapped thresholding areas as vascular regions. (1–4) shows the measurement condition without UCA and time after UCA injection.

Fig. 3 [Images not available. See PDF.]

Visualization of vascular structures in metastatic LN. (A–C) show the B-mode without clutter filter, TIAE map, and overlapped thresholding areas as vascular regions. (1–4) shows the measurement condition without UCA and time after UCA injection.

The vascular area against LN regions, i.e., the vascular area ratio, was quantified in the TIAE image as shown in Figs. 2(C) and 3(C), and confirmed in different conditions without and with UCA being dependent on the time after injection. Figure 4 shows the calculated vascular area ratio in the individual LN. Vascular area ratio in the LN region naturally increased around 10% with UCA compared to without UCA. In addition, the vascular area ratio decreased around 1–3% in time after UCA injection from 0 to 2 min. Especially at 2 min after UCA injection, the vascular area ratio in the metastatic group was lower being about 4% in the control group on average (except for one case), reflecting rapid contrast washout and potential differences in focal defects. It is noteworthy that even before UCA injection (i.e., without UCA), the vascular area ratio differed markedly between the control and metastatic groups. This result indicates that the baseline vascular characteristics alone provide sufficient discrimination between the two conditions.

Fig. 4 [Images not available. See PDF.]

Quantitative evaluation of vascular area ratio in LN regions.

We assessed luciferase activity. In Fig. 5, the change of luciferase activity in days after inoculation was tracked in tumor-bearing mice (Figs. 5(A) and 5(B)). Luciferase activity increased especially on day 8 except for PALN in MXH54/Mo-lpr/lpr (MXH54) mouse (Fig. 5(C)). Metastasis in PALN (Fig. 5(D)) was confirmed on day 18 of C3H/HeJ-lpr/lpr (C3H) and day 35 of MXH54 mice because ex vivo luciferase intensity revealing simultaneous metastases in the lung, suggesting systemic dissemination.

Fig. 5 [Images not available. See PDF.]

In vivo and ex vivo assessment of LN metastasis. (A) Experimental schedule. (B) Bioluminescent images of C3H and MXH54 mice. (C) Time-course of luciferase activity. (D) Ex vivo imaging confirming metastases in PALN and lung in MXH54 at day 35.

Figure 6 compares the UCA imaging (TIAE map) and histopathology to focus on the local features such as focal defects in the microcirculation. Clutter-filtered UCA images revealed the low echogenicity and scattered distributed UCA regions in the local PALN (Figs. 6(A-1) and 6(B-1)). In these regions, tumor nests were observed in histological sections (Figs. 6(A-2), 6(A-3), 6(B-2) and 6(B-3)), which were diagnosed as focal defects.

Fig. 6 [Images not available. See PDF.]

Qualitative comparison of focal defects by ultrasound and histological images. (A-1) C3H mouse shows around 1 × 1 mm² defects; (B-1) MXH54 mouse shows around 2 mm axial × 4 mm lateral defect comparable to the size of tumor in histopathological images (A-2, A-3, B-2 and B-3). Arrows indicate ultrasound-detected tumors; histology confirms tumor nests (T), CT contrast agent (CT-CA), and cortex (Cor).

Discussion

The results demonstrate that high-frequency contrast-free and contrast-enhanced ultrasound, combined with SVD-based clutter filtering offers a method to visualize the LN microcirculation as shown in the Supplementary Movie. The clutter filter process successfully eliminated both tissue signals and cardiac motion owing to the selected singular values as shown in Fig. 7. Each region in Fig. 7 can be explained by the property of SVD that decomposes spatiotemporal data into orthogonal modes in order of energy^14,15. The upper band, low-correlation zone reflects sharp localized motion, while the adjacent high-correlation zone indicates spatially coherent, large-scale organized motion such as tissue clutter. The lower band contains low‑energy, spatially fragmented microvascular flow and noise distributed across many modes. Retaining the lower band for reconstruction preserves these low‑energy vascular components and yields the results shown in the Supplementary Movie. While the small out-of-plane motion would be present, we believe that the TIAE image such as shown in Figs. 2(B) and 3(B) could trace UCAs in typical vascular structure while suppressing the impact of a rapid change in amplitude such as system noise and motion artifact. If the data acquisition frame were smaller, e.g. during a cardiac cycle, the circulation was insufficient in the present study. Moreover, SVD would suffer from distinguishing the boundary between tissue and vascular components in small frames. A high-frame-rate ultrasound imaging^20,21and motion compensation algorithm^22,23 might be useful to enhance the visibility and spatial connectivity of vascular structure. Nevertheless, a key novelty of this study was considered to be that focal defects in metastatic lymph nodes can potentially be detected using contrast-free high-frequency ultrasound without relying on high-frame-rate imaging as evaluated in Fig. 4 (vascular area ratio in the LN).

Fig. 7 [Images not available. See PDF.]

Adaptive SVD-based clutter filter of ultrasound signals. (A) and (B) show the correlation matrix and its enlarged image. (C) Determination of adaptive threshold based on the maximum value of cumulative correlation coefficient.

In the summation image as shown in Figs. 2(C) and 3(C), we compared vascular detectability without or with UCAs. The TIAE map offered noise-robust images optimal for quantification, while the frame-by-frame and MIP methods might suffer from system noise and motion artifact in the speckle pattern. These complementary properties are qualitatively evidenced in Figs. 2 and 3.

In this study, the vascular structural value was quantified using 2D-slice ultrasound, which enables non-invasive, in vivo assessment and allows for longitudinal monitoring of vascular changes, and involves no exposure to ionizing radiation. However, its spatial resolution and volumetric accuracy are limited due to an optionally selected 2D cross-section. One approach to address this limitation is evaluation using 3D ultrasound reconstruction²⁴. However, motion artifacts and operator dependency must still be considered. Future validation against high-resolution ex vivo X-ray-based 3D micro-CT, which provides comprehensive visualization of the vascular network¹⁷, would serve as a robust reference to assess the accuracy and potential bias of ultrasound-based measurements²⁵. In addition, microbubble-based phase-contrast imaging (PCI) provides high spatial resolution and sensitivity to microstructural changes²⁶, and in vivo comparison with radiation-based PCI using similar-sized contrast agents could further validate ultrasound-derived measurements.

Importantly, the biological validity of the imaging approach was corroborated as shown in Fig. 5. The progression from localized PALN metastasis to distant lung involvement in MXH54 mice (Fig. 5(D)) reflects both lymphatic and hematogenous dissemination, making this a valuable preclinical model for multistage metastasis. The detection of metastatic lesions which are micro-sized focal defects within lymph nodes via clutter-filtered ultrasound (Figs. 6(A-1) and 6(B-1)) and their correspondence with histology (Figs. 6(A-2), 6(A-3), 6(B-2) and 6(B-3)) demonstrates not only high spatial accuracy but also its non-invasive monitoring potential. The SVD technique, which forms the basis of our proposed method, has been shown to be effective in evaluating mouse tumor by improving microvascular visualization²⁷. Our approach visualizes lesions even in LNs with challenges such as small size, dense capsule, and internal heterogeneity, suggesting this technique may be useful for diagnosis in metastatic LNs.

Overall, this integrative methodology – merging spatiotemporal ultrasound analysis with histological validation – advances our ability to visualize, quantify and interpret microvascular and metastatic changes in lymphatic organs. Further enhancement would be important for transducer sensitivity, reducing motion artifact, and super-resolution imaging, e.g., point spread function deconvolution, improving the algorithm for isolated vascular extraction, and for validation across larger sample cohorts to confirm clinical translation feasibility.

In the present study, high-frequency ultrasound at 40 MHz was used to visualize the microcirculation without or with UCA. The clutter- filtered images might show the single UCA without collapsing as shown in Supplementary Movies 1 and 2. In UCA imaging using microbubbles, the resonant frequency is around 1-10 MHz²⁸. While harmonic imaging is used in this frequency range, high-frequency ultrasound such as 40 MHz would mismatch when considering the resonant frequency against the size of the microbubble. If the UCA were smaller such as nanobubbles²⁹, high-frequency ultrasound might be reasonable for more accurate UCA imaging. Also, high-frequency ultrasound is one of the clinical applications for in situ measurement to classify the non-metastasis or metastasis LN during surgery for LN dissection.

Methods

Experiments were carried out in accordance with approved guidelines and were approved by the Institutional Animal Care and Use Committee of Tohoku University. This study is reported in accordance with the ARRIVE guidelines (https://arriveguidelines.org).

Mice

MXH54/Mo-lpr/lpr (MXH54) and C3H/HeJ-lpr/lpr (C3H) mice were used in the study³⁰. The MXH54 strain is a recombinant inbred line that spontaneously develops autoimmune disease, derived from MRL/MpJ-lpr/lpr and C3H progenitors. Notably, MXH54 mice exhibit LN enlargement comparable to that observed in MRL/lpr mice. In the experiments, MXH54 were used as the control group (n = 5) and both MXH54 and C3H were used in the metastatic group (n = 2).

Metastasis induction

LM8-Luc cells³¹ were used in the present study. Prior to experimentation, the absence of mycoplasma contamination was confirmed using a Mycoplasma Detection Kit (R&D Systems Inc., Minneapolis, MN, USA). LM8-Luc cells were then suspended in PBS (Sigma-Aldrich) at a concentration of 1.0 × 10⁶ cells/mL. The suspension was mixed with Matrigel (Collaborative Biomedical Products, Bedford, MA, USA) at a 1:2 ratio (v/v), resulting in a final concentration of 3.3 × 10⁵ cells/mL per 60 µL injection. For tumor cell inoculation, mice were anesthetized with 2% isoflurane in air (Pfizer Japan Inc.), and the cell suspension was administered as a bolus into the right subiliac LN (SiLN) using a 27-gauge Nipro Myjector syringe (Nipro Co., Osaka, Japan). To facilitate Matrigel solidification, the needle was left subcutaneously in place for 1 min post-injection, followed by a 180° rotation during withdrawal. Bioluminescence imaging was performed every other day using the method described below, and metastasis was induced until the signal intensity in the proper axillary LN (PALN) reached around 10⁶ photons/sec.

Tumor growth evaluation

Tumor growth was monitored using a bioluminescence imaging system (IVIS Lumina Series III, PerkinElmer, MA, USA). For imaging, mice were anesthetized with 2% isoflurane and positioned on a heated stage maintained at 40 °C to maintain body temperature. D-Luciferin potassium salt (15 mg/mL, FujiFilm Wako Pure Chemical Industries, Ltd.) was dissolved in PBS and administered via an intraperitoneal injection at a dose of 10 µL/g body weight. Bioluminescent images were acquired 10 min after luciferin administration using an exposure time of 1 min. Imaging parameters were set as follows: binning, medium; F/Stop, 1; field of view, 12.5 cm; and subject height, 1.5 cm. Signal intensity was quantified as total flux (photons/sec) using Living Image software (version 4.7.3, PerkinElmer) (Fig. 1(A)).

Ultrasound RF data acquisition

A modified animal ultrasound imaging system (Vevo 770, FujiFilm VisualSonics) equipped with 40 MHz transducer (RMV704, axial resolution ~ 40 µm, lateral ~ 100 µm, FujiFilm VisualSonics) was used¹⁶. This spatial resolution can show the full width at half-maximum. The transducer was mounted on a 3D stage control system (Mark-204-MS, Sigma-Koki, Tokyo, Japan). Mice were anesthetized with 2% isoflurane (Pfizer Japan Inc., Tokyo, Japan) in air and fixed in place with their fur shaved. The transducer was applied to the right proper axillary lymph node (PALN), and the focus was adjusted on the ultrasound monitor to center on the LN. Next, ultrasound RF data was acquired using SBench 6 (MISH International Co., Ltd., Japan) and dedicated measurement software for an A/D board by SPECTRUM. The RF data was collected at a sampling frequency of 250 MHz, and quantized at 14 bits. The pitch of scanline was 50 µm. In-phase/Quadrature-phase (IQ) signal was obtained by the Hilbert transform. The frame rate of an acquired image was 70 (control) and 50 (metastasis group) fps, and total 319 and 189 frames were obtained. The reason why the frame rate differed between the control and metastatic groups was related to the larger LN volume in metastatic group. First, the data were obtained before UCA injection. In UCA imaging, 100 µL of UCA (Sonazoid, GE Healthcare Co., Tokyo, Japan) suspended following the manufacturer’s manual, was injected into the tail vein of a mouse. The data at 0–4 min after UCA injection³², when the UCA circulation and its time intensity curve were stabilized, were acquired as contrast-enhanced ultrasound RF data.

Clutter filter based on SVD

IQ data were reconstructed into a spatiotemporal matrix for input to the SVD-based clutter filter^14,33. Note that because filtering was performed here on the interior of the LN, the inside LN was manually masked. The frames in cardiac and breathing motion, when showing amplitude fluctuations with peak width (20%-rise and -fall width) in Fig. 1(B-2), were applied with zero-padding. Let be the IQ data comprising one frame of the echo image. To this was added the slow-time variable to make the spatio-temporal matrix where denote depth, azimuth and time (image frames). To extract the singular value diagonal matrix by SVD, it was necessary to construct a two-dimensional matrix consisting of spatial and temporal information. Therefore, was reconstructed from 3D to 2D . It could then be decomposed into three matrices as . Note that indicates a transposition operation and indicates a diagonal matrix of the singular value.

1

We applied the spatial singular vector obtained by SVD for the RF signal to the following equation.

2

3

In this context, and represent the spatial distributions corresponding to the i-th and j-th singular values, respectively. The covariance matrix quantifies the similarity between these distributions. Regarding the correlation coefficient matrix , a low threshold of the singular value was selected for discriminating clutter from blood flow components^34,35. We adaptively selected rank of threshold at the maximum correlation coefficient as shown in Fig. 7.

Post-processing

The filtered IQ data were used for its absolute value, i.e., amplitude envelope, and was visualized. In addition to single-frame of amplitude images, time-integrated amplitude envelope (TIAE) were also generated to evaluate the UCA areas. The formulas for TIAE are presented in Eqs. (4). The temporal period was selected during 3.78 s after data collection where was temporal period (1/frame rate). In this equation, denotes the spatial amplitude envelope at each frame.

4

Quantification of vascular structural features

To quantify vascular structures, from the TIAE images, we computed the ratio of vascular and LN areas. To ensure the LN region, a margin was applied to the manually segmented LN mask using morphological erosion with a kernel size of [20 (depth) × 10 (lateral)] pixels. As a preprocessing step, the UCA region was detected by applying a locally adaptive thresholding method in the region of LN, which calculated a Gaussian-weighted local mean over a neighborhood approximately matching the point spread function (PSF) size ([27 (depth) × 7 (lateral)] pixels). A sensitivity coefficient of 0.5 was used to adjust the threshold relative to local intensity variations due to ultrasonic attenuation and tissue heterogeneity. Also, considering the defocus depth of high-frequency ultrasound, the evaluation areas in the LN were limited in the range of depth 6.0 ± 0.75 mm.

Histological analysis

In C3H mice, the PALN was excised on day 18; in MXH54 mice, on day 35. The PALN was immersed in 10% neutral buffered formalin (FujiFilm Wako Pure Chemical Industries, Ltd.) at room temperature for 4 days. Following fixation, the PALNs was dehydrated using 100% ethanol (FujiFilm Wako Pure Chemical Industries, Ltd.). After ethanol replacement, the PALNs were embedded in paraffin using a fully enclosed tissue processor (HistoCore PEARL; Leica Biosystems, Nussloch, Germany). Paraffin sections were cut at a thickness of 2.5 μm. Hematoxylin–Eosin (HE) staining was performed by hand. Pathological images of each organ were acquired using a polarized light microscope (BX51; Olympus) equipped with a digital microscope camera (DP21; Olympus).

Conclusion

This study demonstrated that high-frequency ultrasound and SVD-based clutter filtering effectively separate tissue and vascular structure in LN. While the filtered contrast-free imaging failed to clearly visualize the vascular structures, the interconnected vascular structures were distinctly delineated in the filtered contrast-enhanced imaging. Post-filtering signal processing with TIAE further improved image stability and vascular delineation, supporting the quantitative assessment of microvascular architecture. Moreover, the MXH54 mouse model proved suitable for investigating the complex multi-step metastatic process involving lymphatic and hematogenous dissemination. Importantly, ultrasound imaging with UCA enabled high-accuracy detection of focal defects of the circulation in metastatic tumor and monitoring of tumor progression in LNs, with implications for improving diagnosis and therapeutic evaluation. Future work will address limitations related to imaging artifacts and extend validation to pathological models, advancing this method towards clinical application.

Clinical implications

This approach enables high-precision visualization of LN microvasculature using UCAs, potentially opening new avenues for early detection of micrometastases that have been difficult previously to identify.

Noise reduction and vascular structure enhancement through TIAE techniques hold promise for applications in tumor angiogenesis assessment and treatment monitoring.

Non-invasive, real-time vascular information acquisition may reduce patient burden and improve diagnostic accuracy, supporting clinical adoption.

Future research will focus on correlation with pathological specimens and comparative studies with other contrast agents.

Author contributions

K.M., M.O., A.S., T.S. and T.K.: data curation, formal analysis, investigation, methodology. K.M. and T.K.: writing – original draft. K.M. and M.O.: conceptualization, data curation, formal analysis, investigation, methodology, supervision. A.S., T.S. and T.K.: resources, software, supervision, validation, visualization, and writing – review and editing. T.K.: funding acquisition, project administration.

Funding

This research was supported by JSPS KAKENHI (Grant Numbers: 23H00543, 21K18319, 25K034895 and 25K21824) and KEIRIN RACE (Grant Numbers: 2025 M-329).

Data availability

The datasets used and/or analyzed during the current study are available from the corresponding author on reasonable request.

Declarations

Competing interests

The authors declare no competing interests.

Ethics approval

All animal experiments were approved by the Institutional Animal Care and Use Committee of Tohoku University (approval number: 2019BeA-008).

ARRIVE guidelines

This study is reported in accordance with the ARRIVE guidelines (https://arriveguidelines.org).

Abbreviations

CA

Contrast agent

Cor

Cortex

C3H

C3H/HeJ-lpr/lpr

IQ

In-phase/Quadrature-phase

LN

Lymph node

LDDS

Lymphatic drug delivery system

LV

Lymphatic vessel

MXH54

MXH54/Mo-lpr/lpr

PALN

Proper axillary lymph node

SiLN

Subiliac lymph node

SVD

Singular value decomposition

TIAE

Time-integrated amplitude envelope

UCA

Ultrasound contrast agent

Publisher’s note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

References

1. Wu, Y; Shang, J; Zhang, X; Li, N. Advances in molecular imaging and targeted therapeutics for lymph node metastasis in cancer: A comprehensive review. J. Nanobiotechnol.; 2024; 22, 783. [DOI: https://dx.doi.org/10.1186/s12951-024-02940-4]

2. Bundred, J; Lal, N; Chan, DKH; Buczacki, SJA. Lymph node yield as a surrogate marker for tumour biology and prognosis in colon cancer. Br. J. Cancer; 2025; 132, pp. 643-651. [DOI: https://dx.doi.org/10.1038/s41416-025-02949-y] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/39953281][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11961567]

3. Kodama, T et al. Tumor cell invasion from the marginal sinus into extranodal veins during early-stage lymph node metastasis can be a starting point for hematogenous metastasis. J. Cancer Metastasis Treat; 2018; [DOI: https://dx.doi.org/10.20517/2394-4722.2018.61]

4. Takeda, K; Mori, S; Kodama, T. Study of fluid dynamics reveals direct communications between lymphatic vessels and venous blood vessels at lymph nodes of mice. J. Immunol. Methods; 2017; 445, pp. 1-9.1:CAS:528:DC%2BC2sXjs1OktL8%3D [DOI: https://dx.doi.org/10.1016/j.jim.2017.02.008] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/28237707]

5. Sun, P et al. Predicting axillary lymph node metastasis in breast cancer based on ultrasound radiofrequency time-series analysis. Acta Radiol.; 2024; 65, pp. 1178-1185. [DOI: https://dx.doi.org/10.1177/02841851241268463] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/39219479]

6. Welch, DR; Hurst, DR. Defining the hallmarks of metastasis. Cancer Res.; 2019; 79, pp. 3011-3027.1:CAS:528:DC%2BC1MXhs1eqsLzM [DOI: https://dx.doi.org/10.1158/0008-5472.Can-19-0458] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/31053634][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6571042]

7. Ito, K et al. High-accuracy ultrasound contrast agent detection method for diagnostic ultrasound imaging systems. Ultrasound Med. Biol.; 2015; 41, pp. 3120-3130. [DOI: https://dx.doi.org/10.1016/j.ultrasmedbio.2015.07.032] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/26411669]

8. Shima, H et al. Comparing the extent of breast cancer tumors through contrast-enhanced ultrasound vs B-mode, opposed with pathology: Evergreen study. Breast Cancer; 2021; 28, pp. 405-413. [DOI: https://dx.doi.org/10.1007/s12282-020-01176-y] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/33119842]

9. Chen, Y; Fang, B; Li, H; Huang, L; Luo, J. Ultrafast online clutter filtering for ultrasound microvascular imaging. IEEE Trans. Med. Imaging; 2025; 44, pp. 2477-2491.2025ITMI..44.2477C [DOI: https://dx.doi.org/10.1109/tmi.2025.3535550] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/40031320]

10. Hasegawa, H; Mozumi, M; Omura, M; Nagaoka, R. Improvement of spatial resolution of medical ultrasound images by constrained least-square method. Japanese J. Appl. Phys.; 2021; 60, SDDE16.1:CAS:528:DC%2BB3MXhs1OnsL7K [DOI: https://dx.doi.org/10.35848/1347-4065/abf3a1]

11. Xu, M; Yang, X; Ding, M; Yuchi, M. Spatio-temporally smoothed coherence factor for ultrasound imaging. IEEE Trans. Ultrason. Ferroelectr. Freq. Control; 2014; 61, pp. 182-190.2014ITUFF.61.182X [DOI: https://dx.doi.org/10.1109/tuffc.2014.6689786] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/24402905]

12. Flint, K; Huber, M; Long, J; Trahey, G; Hall, T. Clutter-generating phantom material. part i: development of a tunable, acoustic clutter-generating layer for use with ultrasound tissue-mimicking phantoms. Ultrasound Med. Biol.; 2025; 51, pp. 768-776. [DOI: https://dx.doi.org/10.1016/j.ultrasmedbio.2024.12.018] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/39924418]

13. Ledoux, LA; Brands, PJ; Hoeks, AP. Reduction of the clutter component in Doppler ultrasound signals based on singular value decomposition: A simulation study. Ultrason. Imaging; 1997; 19, pp. 1-18.1:STN:280:DyaK2svjt1SjtA%3D%3D [DOI: https://dx.doi.org/10.1177/016173469701900101] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/9286010]

14. Demené, C et al. Spatiotemporal clutter filtering of ultrafast ultrasound data highly increases doppler and fultrasound sensitivity. IEEE Trans. Med. Imaging; 2015; 34, pp. 2271-2285.2015ITMI..34.2271D [DOI: https://dx.doi.org/10.1109/tmi.2015.2428634] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/25955583]

15. Baranger, J; Aguet, J; Villemain, O. Fast thresholding of SVD clutter filter using the spatial similarity matrix and a sum-table algorithm. IEEE Trans. Ultrason. Ferroelectr. Freq. Control; 2023; 70, pp. 821-830.2023ITUFF.70.821B [DOI: https://dx.doi.org/10.1109/tuffc.2023.3289235] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/37363854]

16. Omura, M et al. Attenuation estimation and acoustic characterization of mouse lymph node tumor using high-frequency ultrasound. Mol. Imaging Biol.; 2025; 27, pp. 379-388.1:CAS:528:DC%2BB2MXisVertrfL [DOI: https://dx.doi.org/10.1007/s11307-025-02007-2] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/40355617][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12162808]

17. Yamaki, T et al. Characterizing perfusion defects in metastatic lymph nodes at an early stage using high-frequency ultrasound and micro-CT imaging. Clin. Exp. Metastasis; 2021; 38, pp. 539-549. [DOI: https://dx.doi.org/10.1007/s10585-021-10127-6] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/34654990]

18. Sakamoto, M et al. Perfusion defects in non-enlarged metastatic lymph nodes using vessel wall magnetic resonance imaging: Detection performance and diagnostic value. Clin. Exp. Metastasis; 2022; 39, pp. 421-431.1:CAS:528:DC%2BB38XhtVWku77K [DOI: https://dx.doi.org/10.1007/s10585-022-10147-w] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/35119560]

19. Xia, S et al. Super-resolution ultrasound imaging of intranodal lymphatic sinuses for predicting sentinel lymph node metastasis in breast cancer: A preliminary study. Eur. Radiol.; 2025; [DOI: https://dx.doi.org/10.1007/s00330-025-11520-5] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/41198863][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12166017]

20. Ketterling, JA et al. High-speed, high-frequency ultrasound, in utero vector-flow imaging of mouse embryos. Sci. Rep.; 2017; 7, 16658.2017NatSR..716658K1:CAS:528:DC%2BC1cXhsFWnsrbP [DOI: https://dx.doi.org/10.1038/s41598-017-16933-x] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/29192281][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC5709407]

21. Omura, M et al. Characterization of blood mimicking fluid with ultrafast ultrasonic and optical image velocimeters. Japanese J. Appl. Phys.; 2022; 61, SG1067.1:CAS:528:DC%2BB2cXktlWksr4%3D [DOI: https://dx.doi.org/10.35848/1347-4065/ac4ea9]

22. Harput, S et al. Two-stage motion correction for super-resolution ultrasound imaging in human lower limb. IEEE Trans. Ultrason. Ferroelectr. Freq. Control; 2018; 65, pp. 803-814.2018ITUFF.65.803H [DOI: https://dx.doi.org/10.1109/tuffc.2018.2824846] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/29733283]

23. Omura, M; Takeuchi, M; Nagaoka, R; Hasegawa, H. Ultrasonic envelope statistical analysis in motion compensated images during temperature change. Japanese J. Appl. Phys.; 2023; 62, SJ1036.1:CAS:528:DC%2BB2cXitVyksLnI [DOI: https://dx.doi.org/10.35848/1347-4065/acbb11]

24. van Sloun, RJG et al. Contrast-enhanced ultrasound tractography for 3D vascular imaging of the prostate. Sci. Rep.; 2018; 8, 14640.2018NatSR..814640V1:CAS:528:DC%2BC1MXpsFOksA%3D%3D [DOI: https://dx.doi.org/10.1038/s41598-018-32982-2] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/30279545][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6168586]

25. Arendt Jensen, J et al. Super-resolution ultrasound imaging using the erythrocytes-part I: Density images. IEEE Trans. Ultrason. Ferroelectr. Freq. Control; 2024; 71, pp. 925-944.2024ITUFF.71.925A [DOI: https://dx.doi.org/10.1109/tuffc.2024.3411711] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/38857145]

26. Tang, R et al. Microbubble-based synchrotron radiation phase contrast imaging: Basic study and angiography applications. Phys. Med. Biol.; 2011; 56, pp. 3503-3512.2011amtm.book...T [DOI: https://dx.doi.org/10.1088/0031-9155/56/12/004] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/21606555]

27. Xia, Y et al. Enhanced SVD filter based on singular vector subspace denoising improves ultrafast ultrasound microvascular imaging performance. Phys. Med. Biol.; 2025; [DOI: https://dx.doi.org/10.1088/1361-6560/ae085f] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/41192056][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC12096871]

28. Hong, YR et al. Role of contrast-enhanced ultrasound in the pre-operative diagnosis of cervical lymph node metastasis in patients with papillary thyroid carcinoma. Ultrasound Med. Biol.; 2017; 43, pp. 2567-2575. [DOI: https://dx.doi.org/10.1016/j.ultrasmedbio.2017.07.010] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/28807450]

29. Wegierak, D et al. Nanobubble contrast enhanced ultrasound imaging: A Review. Wiley Interdiscip. Rev. Nanomed. Nanobiotechnol.; 2024; 16, 1:CAS:528:DC%2BB2cXis12hsLfF [DOI: https://dx.doi.org/10.1002/wnan.2007] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/39511794][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC11567054]e2007.

30. Tanaka, Y et al. Evaluating the role of rheumatoid factors for the development of rheumatoid arthritis in a mouse model with a newly established ELISA system. Tohoku J. Exp. Med.; 2010; 220, pp. 199-206.1:CAS:528:DC%2BC3cXkvVSlu7g%3D [DOI: https://dx.doi.org/10.1620/tjem.220.199] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/20208415]

31. Mishra, R; Sukhbaatar, A; Mori, S; Kodama, T. Metastatic lymph node targeted CTLA4 blockade: A potent intervention for local and distant metastases with minimal ICI-induced pneumonia. J. Exp. Clin. Cancer Res.; 2023; 42, 132.1:CAS:528:DC%2BB3sXht1SisbfK [DOI: https://dx.doi.org/10.1186/s13046-023-02645-w] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/37259163][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC10233978]

32. Spiesecke, P; Neumann, K; Wakonig, K; Lerchbaumer, MH. Contrast-enhanced ultrasound (CEUS) in characterization of inconclusive cervical lymph nodes: A meta-analysis and systematic review. Sci. Rep.; 2022; 12, 7804.2022NatSR.12.7804S1:CAS:528:DC%2BB38XhtlCrsLvI [DOI: https://dx.doi.org/10.1038/s41598-022-11542-9] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/35551228][PubMedCentral: https://www.ncbi.nlm.nih.gov/pmc/articles/PMC9098903]

33. Omura, M; Yagi, K; Nagaoka, R; Hasegawa, H. Contrast analysis in ultrafast ultrasound blood flow imaging of jugular vein. J. Med. Ultrason.; 2023; 2001, 50 pp. 131-141. [DOI: https://dx.doi.org/10.1007/s10396-023-01289-9]

34. Baranger, J et al. Adaptive Spatiotemporal SVD Clutter Filtering for Ultrafast Doppler Imaging Using Similarity of Spatial Singular Vectors | IEEE Journals & Magazine | IEEE Xplore. IEEE Trans. Med. Imaging; 2018; [DOI: https://dx.doi.org/10.1109/tmi.2018.2789499] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/29969408]

35. Omura, M et al. Dual-frequency excitation in high-frame-rate ultrasonic backscatter coefficient analysis of hemorheological properties. Ultrasonics; 2025; 149, 1:CAS:528:DC%2BB2MXjt1OntrY%3D [DOI: https://dx.doi.org/10.1016/j.ultras.2025.107589] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/39914161]107589.