This study was a retrospective study of consecutive patients who underwent navigation-guided brain biopsy using the Stealth Autoguide system performed by five neurosurgeons at three institutions between April and July 2025. Medical records were reviewed to obtain demographic, procedural, and outcome data.

The Stealth Autoguide system is a robot-assisted, automated alignment platform comprising two robotic components : a targeting unit (Fig. 1A) and a control unit (Fig. 1B), both integrated with the navigation system. The targeting unit measures 27.7×11.8×7.6 cm and weighs 1.42 kg, making it compact and lightweight enough to be mounted on a Mayfield head clamp (Fig. 1C).

The surgical workflow is illustrated in Fig. 2. All procedures were conducted under general anesthesia. Also, a drill bit and biopsy needle were single-use disposable instruments, whereas all other components—including the drill guide, sharp obturator, and reference frame—were part of a sterilizable, reusable kit. The patient’s head is secured using a rigid skull clamp, and the targeting unit is mounted contralaterally to the navigation reference frame (Fig. 2A). Target and entry points are selected on the MRI based on optimal trajectory, avoidance of critical vasculature, and surgical accessibility. Following sterile draping, once activated, the robotic system automatically moves to the planned entry point and aligns along the predefined entry-to-target trajectory (Fig. 2B).

Two methods were used to create the scalp entry point and to anchor a drill guide. The first involved making a stab incision less than 1 cm in length using a surgical scalpel, followed by anchoring the serrated drill guide onto the skull. The second method employed a drill guide anchored onto the scalp at the planned entry site, through which a sharp obturator was advanced to puncture the scalp (Fig. 2C). Drilling was performed using a motor-driven twist drill with a diameter of 3.2 mm and an adjustable depth stop in a range of 1-20 mm. Drilling depth was calibrated according to the scalp-to-dura (or skull-to-dura) distance measured along the planned trajectory on preoperative MRI. Percutaneous drilling requires only advancing the drill with appropriately adjusted depth stop along the drill guide, which is already secured to either the scalp or skull. After drilling, biopsy needle insertion and tissue sampling was performed in the standard manner (Fig. 2D).

Depending on the surgeon’s preference and tumor characteristics, the procedure was either concluded immediately after tissue acquisition or deferred until intraoperative frozen section (FS) confirmation. After removing the biopsy needle and associated devices, the puncture site was irrigated with saline to ensure the absence of active cortical bleeding. The wound was then closed using sterile strips or 1-2 skin staples (Supplementary Fig. 1). An immediate postoperative CT scan was obtained to verify accurate targeting and to confirm the absence of hemorrhage.

The target alignment error (TAE) was defined as the three-dimensional (3D) distance between the preoperatively planned target and the needle tip position as displayed on the navigation system interface. TAE was measured immediately before biopsy needle insertion using the navigation system. In brain biopsies, postoperative CT alone often fails to localize the tissue sampling site accurately. Postoperative MRI allows for more precise calculation of the actual TAE; however, due to the retrospective design of this study, postoperative MRI was not obtained in most patients. Accordingly, this study reported the navigation-based TAE as a surrogate for the actual distance between the planned target and the true needle tip position. Operative time was defined as the interval from skin incision to wound closure, including the waiting period for intraoperative FS results.

Descriptive statistics were used to summarize patient characteristics, surgical outcomes, and operative time. Continuous variables were reported as median values with interquartile ranges (IQRs), and categorical variables were presented as counts and percentages.

A total of 17 consecutive patients underwent frameless stereotactic biopsy using the Stealth Autoguide system. Baseline characteristics are summarized in Table 1. The cohort included nine males and eight females, with a median age of 63 years (range, 29-76). Comorbidities included hypertension in nine patients (52.9%) and diabetes mellitus in five patients (29.4%). No patient had chronic kidney disease, liver cirrhosis, coagulopathy, a history of antiplatelet or anticoagulant use, or prior intracranial radiotherapy. Target lesions were in the deep gray matter in four cases (two in the basal ganglia and two in the thalamus), in the subcortical-to-lobar white matter in 12 cases (seven frontal, four parietal, and one temporal), and in the cerebellum in one case. Final histopathological diagnoses included lymphoma in nine patients (52.9%), glioblastoma (GBM), isocitrate dehydrogenase-wildtype in five (29.4%), high-grade glioma other than GBM in one (5.9%), and inflammatory lesions in two (11.8%).

A summary of clinical outcomes is presented in Table 2, with individual patient details provided in Table 3. Stealth Autoguide-assisted biopsy was successfully performed in 16 of 17 patients. In the single failed case (case 3, Supplementary Fig. 2), severe bilateral frontal lobe atrophy likely led to inaccurate sampling. FS revealed only gliotic tissue, prompting conversion to a conventional navigation-guided biopsy system, which subsequently enabled successful biopsy. In three patients, a stab incision was made to anchor the drill guide to the skull directly, and drilling was performed from the skull surface.

The overall median operative time and anesthesia time were 48 minutes (IQR, 33-64) and 93 minutes (IQR, 84-148), respectively. Among cases with intraoperative FS confirmation, the median operative time was 56 minutes (IQR, 42-65) and the median anesthesia time was 99 minutes (IQR, 88-148). In contrast, cases without intraoperative FS confirmation had a median operative time of 38 minutes (IQR, 30-47) and a median anesthesia time of 87 minutes (IQR, 80-90). TAE was measured in 14 of the 16 successful procedures, with a median value of 0.4 mm (IQR, 0.3-0.6). A definitive histopathological diagnosis was achieved in all patients, and no case required revision surgery.

Three cases of asymptomatic hemorrhage were observed. Two involved hematomas along the biopsy tract near the needle entry site (cases no. 7 and 11), while the third (case no. 14), diagnosed with an inflammatory lesion, showed minimal hemorrhage at the tissue sampling site. All three patients remained neurologically intact, and follow-up CT scans confirmed no hemorrhagic progression. One other patient with a lesion involving the motor cortex developed transient hemiparesis, which fully resolved within 24 hours. No cases of infection or procedure-related mortality occurred.

A notable advantage of this system is that it can be implemented without significant modifications to the conventional frameless stereotactic biopsy workflow. Furthermore, the device is compatible with commonly used cranial fixation systems, including stereotactic base frames [20], and with the Medtronic Stealth software, thereby minimizing additional costs and facilitating integration into routine practice.

Although workflow integration was generally smooth, a learning curve was evident not only for the neurosurgeon but also for nursing staff, as some degree of familiarization with the system was required. In our experience, the learning curve was relatively short, with most surgeons achieving proficiency by their third procedure. While some level of experience is necessary for proficiency, the system appears favorable for training, particularly for residents or fellows who are new to stereotactic biopsy and may struggle with precise manual alignment of targets.

The most distinctive feature of the Stealth Autoguide system compared with the conventional technique is that fine-tuning of target alignment is performed with robot assistance in an automated manner rather than by manual adjustment. This feature replaces slow, error-prone manual adjustments, which are considered the rate-limiting factor in conventional procedures, providing fast and mechanically stabilized accuracy that reduces the potential for human error and the need for repeated verification.

Efficiency is further improved using a narrow twist drill, which creates a pinpoint entry and advances along a predefined robotic trajectory. Minimizing the skin incision simplifies the procedure and improves cosmetic outcomes. The advantages of the Stealth Autoguide system are particularly evident in temporal or infratentorial lesions that require transmuscular access, where it can further reduce operative time and soft tissue injury. Importantly, the system is applicable to all indications of conventional frame-based methods, as it automates steps prone to human error and streamlines time-consuming processes without introducing additional limitations.

Both rigid fixation of the drill guide to the skull via a stab incision and docking onto the skin were performed according to the surgeon’s preference. Each method exhibited unique advantages and inherited limitations. Skull fixation enabled the accurate measurement of bone depth from the outer table to the dura, eliminating variability caused by soft tissue compliance and providing more rigid fixation. In contrast, drilling from the scalp further minimized the incision, although it was slightly less stable than rigid fixation. Nevertheless, stability was sufficient to avoid affecting the surgical outcome. In both approaches, the surgical wound was minimal, less than 10 mm in length, and favorable in terms of cosmesis and recovery.

As this study was conducted during the early implementation phase, the surgical team had not yet fully adapted to the device. Step-by-step verification and ongoing discussions regarding workflow optimization contributed to a longer operative time. Nevertheless, the minimum operative time in this series was 34 minutes, even in cases that included waiting for FS results. This suggests that the operative time could be further reduced once the learning curve is overcome. Although all procedures in this study were performed under general anesthesia, a consistent reduction in operative time may allow the procedure to be safely performed under local anesthesia in selected cases.

Stereotactic biopsy using the Stealth Autoguide system was successfully performed in 16 of 17 cases (94.1%). The intraoperative TAE had a median value of 0.4 mm (IQR, 0.3-0.6), enabling accurate targeting even for lesions as small as 5 mm. This level of accuracy is comparable to that reported in previous studies using conventional frame-based systems.

In this study, TAE represents the real-time deviation between the preplanned trajectory and the displayed needle position on the navigation system interface. Although the TAE initially appears as 0.0 mm immediately after the robotic system completes its automatic alignment, it often increases slightly during subsequent steps, such as compressing the drill guide onto the scalp or skull and advancing the needle. These procedural manipulations can introduce minor deviations from the pre-aligned trajectory, as reflected in the updated TAE values displayed on the navigation system. Because TAE reflects the difference between manual and automated robotic alignment, it serves as a practical surrogate for estimating the reduction of operator-dependent error. However, the TAE in this study does not account for absolute targeting error, which would include cumulative inaccuracies from system calibration, image-to-patient registration, and intraoperative brain shift. Despite these limitations, a median TAE of 0.4 mm—within the context of a navigation system that typically has a registration error of 1-3 mm—suggests that the system provides a reasonably comparable degree of accuracy [7]. Due to the retrospective nature of this analysis, TAE measurements were available in 14 cases.

The most significant concern with the Stealth Autoguide system is the lack of direct visualization of the brain cortex during needle insertion. Cortical vessel injury remains possible despite careful planning, as small vessels may not be visible on preoperative navigation MRI or may be overlooked during trajectory design.

Previous studies have raised similar concerns regarding the inability to directly visualize the cortex [15-17]. As a practical precaution, in all 17 cases, we performed saline irrigation immediately after biopsy needle removal by passing saline through the pinpoint needle hole to check for active cortical bleeding. Although the effectiveness of saline irrigation for detecting significant cortical vessel injury remains to be validated, we recommend incorporating this procedure into routine practice. If active bleeding is detected, at least a burr hole around the pinpoint site may be necessary for inspection and hemostasis. However, despite these preventive methods and careful efforts to plan a trajectory that avoids vessels, two cases of cortical hemorrhage occurred, which were not clearly detected during surgery by this irrigation method alone. Considering that both cases were asymptomatic minor bleedings, the hemorrhage was likely caused by an injury to small cortical veins that was not distinctively visible even on MRI. Although the system's inability to coagulate small cortical veins is considered a limitation of the system, a degree of safety can be presumed, as no cases involved clinically significant hemorrhage with neurological deterioration or major vessel injury.

Direct drilling with a motorized twist drill for bone and dura penetration—without prior stab incision—raises concern for potential cortical injury. Although dural penetration is often perceived by the surgeon, and drilling can be halted accordingly, this method is not always reliable, and the risk of cortical violation remains. In this study, both instances of cortical hemorrhage occurred in cases where no stab incision was made, and only a pinpoint scalp opening was created using a mallet. In these cases, compression of the serrated drill guide against the scalp to prevent slippage may have thinned the overlying tissue, resulting in an actual scalp-to-dura distance shorter than the value preoperatively measured on MRI. This discrepancy may have contributed to cortical injury. To mitigate this risk, we recommend creating a 5-10 mm stab incision and securing the drill guide in direct contact with the skull. This technique may provide a safer alternative and reduce the likelihood of drilling-related complications. When scalp penetration is performed using a sharp obturator without a stab incision, the drill’s depth stop should initially be conservatively set to a shorter length, with incremental increases of 1 mm during subsequent drilling if necessary. Also, neuroendoscopic bipolar cautery with a diameter less than 3 mm can be used for bleeding control [4].

First, the current study is limited by its retrospective nature and relatively small number of patients. The number of cases was not adequate to overcome the learning curve for the workflow with the Stealth Autoguide system. Also, our series evaluated the system only for stereotactic biopsy. Since there have been reports of its use in other procedures such as SEEG and LITT, further clinical investigation is warranted to expand its application in the ROK. Finally, although the Stealth Autoguide system can be integrated with the widely used Medtronic Stealth software, institutions that do not employ this platform for navigation would be burdened with the additional cost of acquiring the system.