IOUS
Since its introduction into neurosurgery in the 1980s, IOUS has undergone a developmental trajectory from low-frequency, low-resolution imaging toward high-frequency, high-resolution imaging. In the early stages of ultrasound technology development, two-dimensional grayscale ultrasound predominated; during this period, image quality was suboptimal, resolution was relatively low, and artifacts were prominent.4,5 This technology was primarily employed for deep lesion localization and puncture guidance; however, limited image quality restricted its ability to delineate detailed tumor characteristics.
With continuous advancements in probe technology, the application of high-frequency linear array probes (e.g., 7–15 MHz) significantly enhanced spatial resolution, facilitating more precise observation of brain tissue architecture, tumor boundaries, morphology, internal echogenicity patterns, and their relationships with surrounding structures.2 This proved particularly valuable for intraoperative localization of infiltrative tumors such as gliomas.6 During IOUS procedures, two-dimensional ultrasound is first utilized, followed by the integration of other ultrasound techniques to acquire more comprehensive multimodal information encompassing anatomical and functional imaging characteristics.5
In recent years, the integration of technologies such as elastography and Doppler flow imaging has further expanded the capabilities of IOUS, enabling it to provide not only anatomical information but also insights into tissue stiffness and hemodynamic status.3,9 Ultrasound elastography, as an emerging non-invasive imaging technique, facilitates tumor identification through quantitative or semi-quantitative assessment of tissue stiffness differences.10–12 Currently, elastography techniques commonly employed in brain tumor surgery include strain elastography and shear wave elastography (SWE).13,14 The former, a quasi-static strain imaging modality, generates tissue strain by applying stress to the region of interest, allowing assessment of tissue softness or hardness based on ultrasound echo signals; the latter, a dynamic elastography technique, utilizes color-coded mapping for differentiation and offers superior reproducibility and contrast compared with strain elastography.
Chauvet et al.15 demonstrated in their 2016 study that differences in stiffness exist between low-grade and high-grade gliomas (HGGs), as well as between low-grade gliomas (LGGs) and normal brain tissue. In the study by Wang et al.,16 SWE was employed alone during brain tumor surgery, with postoperative magnetic resonance imaging (MRI) serving as the gold standard for assessing residual tumor detection accuracy. The study found that SWE yielded a Kappa value of 0.714 with MRI, comparable to the 0.717 achieved by two-dimensional ultrasound, suggesting comparable performance in identifying residual tumor. Cepeda et al.17 applied strain elastography to 102 patients with brain tumors, calculating mean tissue elasticity values through semi-quantitative analysis. They identified significant differences in elasticity values among tumors of different pathological types (P < 0.001). A substantial body of research indicates that elastography provides valuable supplementary information for surgeons during cranial tumor procedures.14,15,17,18
CDFI utilizes Doppler technology to visualize blood flow direction and measure flow velocity. In the context of brain tumor surgery, it primarily encompasses color Doppler ultrasound and power Doppler ultrasound.19,20 In imaging results, red typically indicates blood flow moving toward the transducer, whereas blue denotes flow moving away from the transducer.5 The optimal timing for this technique is after craniotomy but before dural incision. This technology provides comprehensive information regarding tumor vascularity,3 including the relationship between major vessels and the tumor, tumor vessel caliber, tumor feeding arteries, and tumor-draining veins. Such information assists surgeons in more precisely determining surgical location and depth while avoiding injury to critical vessels. However, color Doppler is notably susceptible to scanning angle effects,4,19 and image quality degrades with deeper target locations. Large sampling windows may introduce artifacts; while it demonstrates advantages in visualizing larger vessels, its capacity for microvascular visualization is limited.21 In contrast, power Doppler is unaffected by flow angle and direction,2 offering greater sensitivity in detecting small-caliber and deep-seated vessels. Nevertheless, power Doppler does not provide information on flow direction. Previous studies have demonstrated that Doppler ultrasound assessment of critical tumor and peritumoral vessels reduces intraoperative vascular injury and decreases postoperative complications.3,19
MFI and superb microvascular imaging (SMI), as emerging Doppler imaging techniques,21–25 effectively suppress tissue motion artifacts, offer high resolution, and preserve microvascular blood flow signals. Their capability in displaying low-velocity blood flow surpasses that of conventional color and power Doppler. The terminological distinction primarily reflects vendor-specific nomenclature rather than fundamental technical differences.3,26 However, reports on their application in cranial surgery remain relatively scarce.
In 2017, Ishikawa et al.21 first applied MFI in cranial surgery, describing the MFI characteristics of various brain tumors and proposing that MFI could be used to differentiate tumors from peritumoral tissues. In 2023, the same team26 applied MFI in 20 patients diagnosed with brain tumors. Their findings revealed that when MFI was performed alone, it was difficult to clearly visualize blood flow in numerous small vessels; however, after the administration of a contrast agent, repeated MFI examinations achieved clear visualization. Consequently, they concluded that contrast agent injection enhances the imaging performance of MFI, enabling clear depiction of tumor microvascular architecture, and that analysis of vessel density and contrast agent arrival time allows assessment of tumor blood supply characteristics.
In 2024, Cai et al.22 demonstrated that SMI significantly outperformed grayscale ultrasound in improving the delineation of HGG boundaries (P = 0.033). This study directly compared the clinical value of CDFI and SMI, confirming the advantages of SMI as an advanced technology over CDFI. In 2025, Dixon et al.27 employed SMI to acquire tumor microvascular images and conducted both qualitative and quantitative analyses, concluding that HGG exhibits greater vascular complexity and disorganization, thereby establishing SMI as a novel intraoperative real-time tool for glioma grading.
Currently, MFI still presents several limitations, including restricted visualization of deep-seated lesions and a lack of standardized quantitative parameters.4 Future applications in cranial surgery are expected to involve its combination with functional techniques such as CEUS to provide surgeons with more comprehensive information.
Navigated IOUS integrates ultrasound probes with neuronavigation systems to achieve real-time imaging during surgery, which can be compared with preoperative MRI. This approach effectively addresses the localization challenges posed by brain shift that affect conventional ultrasound. This technology combines the real-time convenience of ultrasound with the spatial localization advantages of navigation.28,29 In 2025, a study by Cepeda et al.30 validated a deep-learning model for glioma segmentation on 2D IOUS images from 197 patients. Concurrently, the team employed the YOLO11 architecture on 1,732 IOUS images,31 achieving outstanding performance, where the mean average precision at an IoU threshold of 0.50 (mAP@50) was 0.95, and the mAP averaged across IoU thresholds from 0.50 to 0.95 (mAP@50-95) was 0.65 at a processing speed of 34 frames per second. Neurosurgeons confirmed that the system integrated seamlessly into the surgical workflow, providing real-time, accurate prediction and delineation of tumor regions. These findings validate the clinical feasibility of AI-assisted IOUS. Vahdani et al.32 proposed a novel deep learning architecture named U-ConvNext, specifically designed for IOUS segmentation of LGGs. This study was the first to introduce uncertainty quantification into IOUS segmentation, providing a reliability assessment for AI-assisted decision-making and holding significant clinical translational value. Zeineldin et al.33 developed NeuroIGN, an open-source, multimodal, explainable AI neuronavigation system that seamlessly fuses preoperative MRI with real-time IOUS, offering intuitive decision support for intraoperative navigation. The advancement of navigated IOUS and AI signifies the transition of IOUS from an auxiliary localization tool to a multifunctional intraoperative navigation platform. It is anticipated that soon this technology will become a cost-effective and precise intraoperative guidance tool in precision neurosurgery.
Despite the promising performance of YOLO11,31 U-ConvNext,32 and NeuroIGN in IOUS image analysis,33 several barriers hinder their clinical adoption, namely the scarcity of large-scale, standardized, multicenter annotated datasets, limited cross-device and cross-scenario generalizability, and high computational requirements for real-time processing.4,31
CEUS
CEUS involves intravenous administration of ultrasound contrast agents (primarily gas-filled phospholipid microbubbles, such as those containing sulfur hexafluoride and perflutren lipid microspheres), which significantly enhances blood contrast and enables dynamic visualization of tumor microvascular kinetics.7,34 These microbubbles exhibit excellent biocompatibility, remain confined to the vascular space without extravasation, undergo no metabolism in vivo, and are eliminated via the pulmonary circulation, offering a high safety profile with no nephrotoxicity or hepatotoxicity.35 Absolute contraindications include known hypersensitivity to sulfur hexafluoride or perflutren components. Caution is advised in patients with severe cardiopulmonary disease or the presence of right-to-left shunts, as microbubbles may enter the systemic circulation.35 The underlying principle is that microbubbles undergo nonlinear oscillation under ultrasound exposure, generating harmonic signals that facilitate high-contrast, high-resolution imaging under low mechanical index conditions.34 Consequently, CEUS provides clear visualization of intravascular blood flow distribution, unaffected by scanning angle, and enables real-time dynamic display of comprehensive vascular information within brain tumor tissues, including anatomical details, tumor perfusion characteristics, and adjacent vessel information, thereby assisting surgeons in intraoperative tumor boundary assessment and tumor grading.5,35
In clinical practice, administration and dosage typically consist of an intravenous bolus injection of 1.2–2.4 mL of contrast agent, followed by a 5–10 mL saline flush. Repeat administrations (usually 2–3 times) are safe.35 For perfusion analysis, continuous infusion is recommended.35 CEUS imaging should be performed at three key time points: (1) after dural opening (baseline), (2) during resection (residual tumor detection), and (3) after resection (final check). Each acquisition should last 60–120 seconds to capture the arterial phase (10–30 s), parenchymal phase (30–60 s), and venous phase (60–120 s).4,35 A low mechanical index (<0.2) should be used to preserve microbubble integrity.35Figure 1 provides a schematic overview of the intraoperative CEUS operational steps and key parameters.4,35
The application of CEUS in cranial settings dates to 1993, when Bogdahn first utilized it for cerebral vascular examination.36 To date, CEUS has achieved significant advancements in cranial surgery, diagnostics, and therapeutics,37 with applications including glioma grading,38 brain tumor surgery,3 diagnosis and evaluation of cerebrovascular diseases,39,40 and even microbubble-mediated enhancement of targeted therapeutic efficacy.41,42 In brain tumor surgery, CEUS has been employed to differentiate residual lesions, protect vascular structures, and distinguish tumor boundaries from peritumoral edema.2
The study by Wang et al.16 demonstrated that CEUS exhibited the highest diagnostic concordance with MRI, outperforming B-mode ultrasound, MFI, and SWE, strongly supporting CEUS as the preferred technique for intraoperative assessment of residual lesions. CEUS enhancement patterns correlate with tumor pathology and angiogenic characteristics. Cheng et al.38 confirmed that HGGs typically display rapid, heterogeneous “ring-like” enhancement with non-enhancing central necrotic areas and highly enhancing peripheral viable tumor tissue, reflecting pathological features of dense microvasculature at the tumor periphery and central ischemia-necrosis. LGGs manifest as slow, homogeneous, or mild enhancement, with some tumors exhibiting no enhancement due to low microvascular density, suggesting that CEUS may have limitations in LGG identification and that multimodal imaging approaches are necessary. Meningiomas, as highly vascular tumors, commonly exhibit homogeneous, marked enhancement on CEUS, with sharp tumor–brain interfaces.5 CEUS features of metastatic tumors are influenced by the vascular characteristics of their primaries, with most hypervascular metastases demonstrating rapid, marked enhancement and relatively well-defined margins.43
Notably, ultrasound contrast microbubbles remain intravascular and do not ordinarily extravasate into brain parenchyma; therefore, CEUS enhancement in brain tumors mainly reflects abnormal tumor microvascularity and perfusion rather than direct microbubble entry into the parenchyma.35,38 Furthermore, microbubbles can serve as therapeutic vehicles; upon ultrasound triggering, they can transiently open the BBB, facilitating targeted delivery of chemotherapeutic agents, antibodies, and immunomodulators, thereby enabling theranostic integration.35,41,42 Thus, CEUS not only addresses the limitations of conventional IOUS in perfusion quantification and boundary identification but also expands its potential for precision therapeutic applications. From a clinical workflow perspective, CEUS requires no additional hardware beyond standard ultrasound equipment, apart from the intravenous administration of a contrast agent. This low-barrier adoption, combined with its real-time imaging capability, positions CEUS for widespread integration into neurosurgical practice.
In summary, conventional ultrasound offers multiple advantages in brain tumor surgery, including ease of operation, low cost, and excellent repeatability. However, it exhibits significant limitations in identifying residual tumors, delineating vascular structures, distinguishing tumor boundaries from peritumoral edema, and assessing the completeness of postoperative resection, particularly when tumor echogenicity differences are subtle, which may lead to missed residual lesions.6 Furthermore, conventional grayscale ultrasound cannot accurately assess tumor microcirculatory perfusion status, creating an information gap for IOUS-guided precise resection.4,7 CEUS precisely fills this gap. By intravenously administering pure blood-pool microbubble contrast agents, CEUS enables dynamic visualization of tumor microvascular perfusion patterns.7,35 Compared with surrounding normal brain tissue and peritumoral edema, glioma tissue generally shows stronger enhancement on intraoperative CEUS, with HGGs demonstrating earlier enhancement than adjacent edema and normal brain tissue.38 This functional contrast makes CEUS a valuable tool for identifying viable tumor boundaries and residual lesions, thereby complementing the structural imaging provided by conventional IOUS (Table 1).2–4,13,15–17,21,35,38
Table 1Comparison of advantages and limitations of different intraoperative ultrasound modalities in brain tumor surgery
| Modality | Imaging principle | Key advantages | Major limitations | Refs |
|---|
| Conventional IOUS | Structural imaging based on differences in acoustic impedance between tissues | Real-time, convenient, low-cost; Accurate localization and puncture guidance | Boundary interpretation is subjective and echo-dependent; Unable to assess microcirculatory perfusion | 2,4 |
| CEUS | Pure blood pool imaging utilizing nonlinear harmonic signals of microbubbles under low acoustic pressure | Functional imaging: visualizes tumor microcirculatory perfusion; Accurately delineates viable tumor boundaries and residual lesions | Requires intravenous contrast agent injection; LGG may show iso-enhancement | 2,16,35,38 |
| SMI | Adaptive algorithm suppresses tissue motion artifacts to visualize slow blood flow | Contrast-agent-free visualization of microvascular architecture; Improves HGG boundary delineation | Reduced resolution for deep-seated lesions; Unable to provide perfusion information | 3,4,21 |
| SWE | Quantitatively calculates tissue stiffness (Young’s modulus) by measuring shear wave velocity | Provides objective quantitative stiffness values; Assists in tumor grading and infiltrative zone assessment | Susceptible to intraoperative interference (e.g., electrocautery); Lacks standardized universal cutoff values | 4,13,15,17 |