A systematic search was performed on 8 VIII 2024. We have found 228 articles from PubMed, 593 from Embase, 38 from Scopus, 36 from Cochrane Reviews, and 362 from Web of Science. A total of 1257 papers were imported into ZOTERO bibliography manager. Bibliographic software identified 454 duplicate articles, which were removed before the screening process. A total of 803 articles underwent preliminary screening of the papers by two authors of this paper independently. We have excluded 759 articles, which were not relevant to the topic of this synthesis. All papers eligible for full-text screening were retrieved. Forty-four papers underwent inclusion screening process by two authors independently. We have excluded 17 papers without hydrocephalus cases and performing RA intervention on this pathology, 2 abstract-only papers, 1 non-human study, 2 studies where RA was not used for the surgical intervention, 6 review papers, 2 studies in Chinese, 1 non-original study, 1 protocol document, and 1 study with duplicate results. Therefore, we found 11 studies from the database search.
We have also performed a manual search through references and citations, and we have found one more study, making a total of 12 papers included in this systematic review [15–26]. Figure 1 shows detailed screening process with PRISMA flow diagram.
A total of 12 studies were included in the systematic synthesis. We have included papers from USA, Germany, France, Taiwan, China, Italy, Austria, and Japan. Studies varied in number of included patients, however no study exceeded more than 20 patients treated with RA. Common reported symptoms were headache, vision disturbance, gait disturbance, and memory problems. Causes of the hydrocephalus also varied between the studies, and included craniopharyngioma, hematoma, pineoblastoma, glioblastoma, meningitis, aneurysms, and aqueductal stenosis. Different robotic workstations were applied in the papers, including ROSA (Zimmer Biomet), NaoTrac (Brain Navi), Evolution 1 (U.R.S. Universal Robot Systems), SurgiScope, Remebot, and NeuRobot (Hitachi).
Some studies included complications such as unsuccessful ETV procedure or minor bleeding, however no major complications occurred with RA. Procedure time varied between included studies, as studies reported total procedure time from different stages of the surgery. Detailed information about the patients and procedures is available in Table 1 & Table 2.
Abbreviations for table: ICA internal cerebral artery, SAH subarachnoid hemorrhage, IVH intraventricular hemorrhage, SDH subdural hematoma, TBI traumatic brain injury, ICH intracerebral hemorrhage, ICP intracranial pressure
3.1. Procedures in included studies
One of the very first attempts of robotic interventions in hydrocephalus was performed by Zimmermann et al. in 2002 and 2004 with the ‘Evolution 1’’ (U.R.S. Universal Robot Systems, Schwerin, Germany) robotic system [20, 21]. A total of six patients underwent endoscopic third ventriculostomy [20]. Before the procedure was performed, Zimmermann and his colleagues needed 30 to 60 minutes for registration of the patient and setup of robotic system. The endoscopic part of the intervention took between 17 to 35 min. In five cases, RA ETV was performed successfully. However, in one case the surgical team experienced robotic system failure, and manual conversion was needed. Nevertheless, no bleeding or other complications occurred in this study, marking this trial as pioneering.
Liu et al. have not only performed the biggest trial of VPS shunting with RA, but also were the only authors, who have compared robotic intervention to manual approach [24]. A total of sixty patients with hydrocephalus were recruited in the study. Twenty of them underwent surgical procedure with Remebot robot, while other forty were treated with traditional methods.
In Remebot group, CT were uploaded into the robotic software, where 3D segmentations of the ventricular region & CSF fluid were performed. Patients were placed in supine position, and their heads were held with Mayfield head holder. Videometric tracker (MicronTracker, ClaroNav, Toronto, Canada) was installed above patient’s head, and registration of the patient was performed. Then entry point was marked, area of the head draped, anesthesia injected, incision with burr hole drilling performed, and dura mater was pierced with unipolar electrocautery. Accuracy of the procedures was validated, and dura was perforated. Cannulas with drainage tube were advanced into desired point, and anchored into the skull after vitals were checked.
Robotic group has achieved statistically lower operation duration time (29.75 ± 6.38 min vs. 48.63 ± 6.60 min, p < 0.05), lower intraoperative blood loss (10.0 ± 3.98 ml vs. 22.25 ± 4.52 ml, p < 0.05), higher accurate puncture rate (100% vs. 77.5%, p < 0.05), and smaller burr hole diameter (4.0 ± 0.3 vs. 11.0 ± 0.2 mm, p < 0.05) compared to traditional technique. Only one complication occurred in RA group (bleeding), while in conventional group there were 10 bleeding complications, and 3 infections. Additionally, among 15 patients in conventional group, the drainage tube contacted with choroid plexus, while in RA this has not happened in any of the cases. Notably, hospitalization time has not achieved statistically significant differences between robotic and conventional groups (3.5 ± 2.2 days vs. 4.2 ± 1.5 days, p = 0.12).
Hoshide al. have conducted an endoscopic third ventriculostomy among pediatric patients with ROSA surgical system assistance [19]. Authors have applied both MRI and CT imaging techniques, which were later used for trajectory planning. Temporary entry point was set within the foramen of Monro point, and adjusted with the robotic software. Patient was placed in supine position and ROSA’s laser was used for patient registration. When other procedures, such as cerebral biopsy were performed, the one requiring more precision was firstly performed, as changes in ventricular anatomy could occur. No. 15 blade was used for skin incision, and drilling of burr hole was performed. No. 67 blade was used for incision of dura mater, and ventricular catheter was intracranially inserted. Endoscope was inserted through plastic bushing, and guided into lateral ventricle. As foramen of Monro was identified, endoscope was guided into the third ventricle. Ventriculostomy was performed anterior to mammillary bodies. Bugbee wire was advanced through endoscope, puncture of the ventricle’s floor was performed, and dilation of floor was performed with Fogarty balloon. If the results were satisfactory, surgical equipment was carefully removed, and incision of the skin sutured.
Nine patients had ETV procedures performed with RA. These patients were diagnosed with tectal mass, porencephalic cysts, Chiari I malformation, aqueductal stenosis, posthemorrhagic hydrocephalus, leptomeningeal metastasis, and pineal region tumor. Authors have reported no intraoperative complications caused by robotic guidance, neither bruising nor stretching of the CNS structures. No neurological deficits were observed among the patients. At 30 day follow-up, all patients had successfully performed interventions, as no patient needed rerun of the procedure. One patient died, however this was unrelated to robotic procedures. Authors have noticed a learning curve, as time required for the operation started decreasing, as more operations were performed. Mean surgery time was 109.4 (± 34.2) minutes.
It's worth mentioning, about technical limitations, as two first procedures were not fully performed with RA. While ROSA was used for stereotaxy and drilling, too long steel bushing was too long and endoscope could not reach ventricle. This was quickly resolved with new bushing in next 7 cases.
Alan et al. have performed intraparenchymal right basal ganglia hematoma evacuation and intraventricular catheter placement with the ROSA stereotactic robot [16]. Among the patients, there was one 76 year-old male with hydrocephalus. Volume of hematoma reduced by 83.50% and the hydrocephalus was resolved. Additionally, authors have analyzed the overall mean error with stereotactic CT scans in the study, which revealed to be 3.48 mm ± 2.34 mm.
Jarebi et al. have described a unique, two-step robotic approach for a 38-year-old patient [15]. A man with complaints including headache, memory problems, polyuria, pan-hypopituitarism, and blurred vision was admitted for MRI, which revealed suprasellar mass and hydrocephalus. Results suggested presence of craniopharyngioma. Patient was immediately admitted, and underwent two robotic procedures. Firstly, with the use of ROSA stereotactic robot and bifrontal right approach, the intracystic catheters were installed for double Ommaya reservoir. Patient symptoms were resolved, hydrocephalus significantly reduced, and he was discharged on third post-operation day. Three weeks later robot assisted CyberKnife stereotactic radiosurgery was performed. No complications occurred, and patient resolved his symptoms.
Chiu et al. from Taiwan have tested two novel technologies in their study [18]. Generally, the use of ROSA-like stereotactic robots requires the applications of bone fiducials for patient registration, which are invasive method [27]. Alternative solutions, such as laser facial registration are not very time-effective [28]. Authors have proposed a solution to these two issues by applying machine vision (MV) registration with NaoTrac robotic system (Brain Navi Biotechnology Co., Ltd.), as an alternative approach.
MV is a relatively novel technology, where artificial intelligence (AI) techniques are used for image processing to interpret, predict, and analyze video-based data information [29–31]. MV has already been applied in various diagnostic fields, such as radiology, ophthalmology and dermatology. Recently, MV has been introduced into surgical field, where it proved to be a helpful aid into decision-making [32, 33]. With the use of facial recognition camera and MRI or CT DICOM, nearly half million data points are recorded and used for creation of 3D face image with NaoTrac system.
A frameless catheter for EVD placement interventions were performed in 14 patients, which needed ventriculostomy (NOTE: one patient did not have hydrocephalus, however due to majority of hydrocephalus cases [13/14, 93%], overall results were presented for this study). Most of the patients (12) underwent pre-operative CT imaging on day of the surgery. Patient was placed in the supine position and had their head placed in Mayfield headrest. Scans were then imported into NaoTrac software, and 3D brain structure was created. Machine vision robotic arm registered patient’s face, and hundreds of thousand data points were captured from the face structures. Accuracy was checked before the procedure, and then surgical team proceeded to sterilization process. Surgical instruments were registered by robotic cameras, and their accuracy was also pre-checked before the surgery. Robotic arm was positioned and burr hole was drilled. Then robotic arm guided the catheter precisely into the desired point. Touch screen of the robot allowed for real time analysis of the movement, which was under supervision of the neurosurgeon.
This procedure was successful in first attempt in 13 out of 14 cases. In one hydrocephalus case, due to the specific anatomy of the ventricles (ventricular decompression), surgeon had to place the catheter deeper into the brain structurers. No surgery related complications, such as bleeding or infection were reported in the study. The mean time for patient registration with MV technology was 142.8 seconds. Total procedure time was between 45 to 60 min. Mean target error with robotic assistance was 1.63 mm, and mean angular error was 1.99 degrees.
Furthermore, authors have analyzed, that price and maintenance of NaoTrac is about 880,000 USD for 8 year period. With 60 procedures per year, this cost would be 1830 per case. With over approximately 600 interventions per year, expenses could be decreased to 186 USD. The surgical expenses with this robotic setup are lower with ROSA system, and reach 2900 USD per case in Taiwan.