Robotics and navigation in spine surgery: A narrative review

Abstract

Introduction

In recent decades, there has been a rising trend of spinal surgical interventional techniques, especially Minimally Invasive Spine Surgery (MIS), to improve the quality of life in an effective and safe manner. However, MIS techniques tend to be difficult to adapt and are associated with an increased risk of radiation exposure. This led to the development of ‘computer-assisted surgery’ in 1983, which integrated CT images into spinal procedures evolving into the present day robotic-assisted spine surgery. The authors aim to review the development of spine surgeries and provide an overview of the benefits offered. It includes all the comparative studies available to date.

Methods

The manuscript has been prepared as per “SANRA-a scale for the quality assessment of narrative review articles”. The authors searched Pubmed, Embase, and Scopus using the terms “(((((Robotics) OR (Navigation)) OR (computer assisted)) OR (3D navigation)) OR (Freehand)) OR (O-Arm)) AND (spine surgery)” and 68 articles were included for analysis excluding review articles, meta-analyses, or systematic literature.

Results

The authors noted that 49 out of 68 studies showed increased precision of pedicle screw insertion, 10 out of 19 studies show decreased radiation exposure, 13 studies noted decreased operative time, 4 out of 8 studies showed reduced hospital stay and significant reduction in rates of infections, neurological deficits, the need for revision surgeries, and rates of radiological ASD, with computer-assisted techniques.

Conclusion

Computer-assisted surgeries have better accuracy of pedicle screw insertion, decreased blood loss and operative time, reduced radiation exposure, improved functional outcomes, and lesser complications.

1. Introduction

With an increasing percentage of an aging population, advancements and better access to imaging as well as improved safety in surgical and anaesthesia techniques over the past couple of decades, spinal surgical interventional techniques have been on a rising trend. These interventions have aimed to improve the quality of life in an effective and safe manner.^1,2 In the quest to minimize complications and offer patient-friendly procedures, the trend has progressively shifted from traditional open approaches to minimally invasive surgery (MIS), in which there are smaller incisions, less damage to surrounding soft tissues, decreased blood loss, and shorter hospital stays.³

However, MIS techniques tend to be difficult to adapt, with steep learning curves, especially in distorted anatomical motion segments, leading to superior facet joint violations and injuries to surrounding neural elements, as they become difficult to appreciate. MIS techniques are also associated with an increased risk of radiation exposure.^4,5 These shortcomings gave rise to the concept of ‘computer-assisted surgery’, with the belief that reliance on technology in stereotaxis may decrease error rates.⁶

In this narrative review, the authors aim to review the development of computer-assisted spine surgeries over the years and provide an overview of the benefits offered along with certain drawbacks of these systems, which are hard to overlook. With escalating technological advancements and keeping in mind patient safety, these systems may well replace the traditional freehand (FH) instrumentation techniques at a larger scale. This article compares and reviews all the computer/technological assisted techniques to date.

2. Methods

The manuscript has been prepared to adhere to the quality standards for narrative reviews, as defined and quantified by “SANRA-a scale for the quality assessment of narrative review articles”. The authors searched Pubmed, Embase, and Scopus using the terms " (((((Robotics) OR (Navigation)) OR (computer assisted)) OR (3D navigation)) OR (Freehand)) OR (O-Arm)) AND (spine surgery)". 23,374 articles were located in the aforementioned databases: 4062 in Pubmed, 5566 in Embase, and 13,746 in Scopus (Fig. 1). Articles were considered if a Title/Abstract search was conducted in English and the entire text was either freely available or made available by the authors. In this review, we only included prospective and retrospective comparative research. No review articles, meta-analyses, or systematic literature searches were considered. No distinction was made between the types of Robotic or Navigation systems employed, therefore all relevant studies were included. After eliminating duplicates and reviewing abstracts, 87 studies were selected for full-text analysis. After reading the full texts, 30 articles were disqualified, and 11 more were found by searching the articles' citations. Following article verification, a total of 68 articles were included for analysis. The majority of the articles belonged to China (24/68) followed by the USA (19/68).

Fig. 1.

Flowchart depicting the method of literature search.

3. Results

3.1. History

Computer-assisted surgical technique was first developed by Brown Roberts in 1983, and integrated computer-tomography (CT) images primarily into Neurosurgical procedures.⁶ Subsequently, the first stereotactic navigation-based surgical intervention for the spine was reported by Girardi et al.⁷ in 1999, using the StealthStation (Medtronic, Memphis, Tennessee, USA). The Light emitting diode (LED) fitted bone clamps were attached to a bony structure on the patient and a virtual 3D dataset was registered using pre-operative patient data (CT images). However, this system could not accurately extrapolate the images obtained pre-operatively in the supine position to Intra-operative navigation guided pedicle screw placement. Nevertheless, this limitation was resolved by the introduction of O-arm (Medtronic, Dublin, Ireland) in 2005, which used the cone-beam CT method to intra-operatively scan and integrate better quality images.⁸ Consequently O-Arm II was launched to cover more surface area in a single scan. Thereafter, the U.S. Food and Drug Administration (FDA) approved a similar intra-operative CT – Navigation integrated complex, Airo (Brainlab, Feldkirchen, Germany, 2013) with comparable accuracy as O-arm but with better image quality.⁹ Another 3D C-Arm-based navigation system, the Ziehm Vision RFD 3D (Ziehm Imaging, Orlando, Florida, USA) was approved by US FDA for use in spine surgeries with benefits alike. To minimize errors due to stereotaxis and movement of the registration clamp, in 2018, the 7D Surgical system (7D Surgical Inc., North York, Ontario, Canada) was introduced. It allows for rapid intra-operative re-registration using direct light and surface mapping to integrate optical topographic images.¹⁰

The focus in the 21st century has been to further build computer-assisted technologies and gradually eliminate the drawbacks of navigation systems and human errors along with complementing traditional MIS. Consequently, the first spinal robotic system, the Spine Assist (Mazor Robotics Ltd., Cesarea, Israel), was US FDA approved in 2004, followed by its successor, the Renaissance Guidance System (Mazor Robotics Ltd.) in 2011.¹¹ The software in consort with the navigation system allows the surgeon to predetermine an optimal pedicle screw trajectory and later on, the robotic arm acts as an aid to the instrumentation and placement of planned pedicle screw. However, to eliminate technological errors due to software integrations, the first stand-alone robotic system was launched in 2016, the Mazor X (Mazor Robotics Ltd.) and later on, in 2018, the O-Arm stealth navigation integrated robotic system came into use.¹² The ROSA One Spine (Zimmer Biomet, Montpellier, France), originally FDA approved in 2016, and the Excelsius GPS (Globus Medical, Audubon, PA, USA), which was FDA approved in 2017, are also fully integrated robotic systems.

3.2. Accuracy

3.2.1. CT navigation vs freehand (FH) techniques

Computer tomography-assisted systems were the first iteration of navigation applied to spinal surgery.⁶⁷ CT images provided precise anatomical information in three dimensions, allowing intraoperative guidance. Numerous studies have demonstrated the usefulness of this.^67,68,70, 71, 72, 73^,75, 76, 77^,80 (Table 1) A 91-patient randomized control trial (RCT) was conducted to compare the accuracy of pedicle screw insertion using conventional versus CT-based navigation methods.⁸⁰ The rate of pedicle breach was considerably greater in the conventional group (13.4%) than in the computer-assisted group (4.6%; P = 0.006). In the conventional group, 1.4% of the screw insertions had pedicle perforations greater than 4 mm, while none did so in the computer-assisted group. However, Fu et al.⁷⁰ reported that there was no significant difference in accuracy between the conventional and navigated groups.

Table 1.

Comparison of Pedicle screw accuracy using various techniques. FH – Free Hand, RG – Robotic guided, NV – Navigation, CT – Computer Tomography, 3DFL – Three Dimensional Fluoroscopy.

Author	Year	Study type	Patients	Accuracy			Difference	Robot
				FH	RG	NV
Comparison of pedicle screw accuracy with free hand and robotic guided techniques
Ringel et al.13	2012	Prospective RCT	60	93%	85%	–	Reduced	SpineAssist
Hyun et al.14	2016	Prospective RCT	60	98.60%	100%	–	No Difference	Renaissance
Wang et al.15	2017	Prospective RCT	30	95%	100%	–	Improved	TiRobot
Kim et al.16	2017	Prospective RCT	78	99.50%	99.40%	–	No Difference	Renaissance
Han et al.17	2019	Prospective RCT	234	93.50%	98.70%	–	Improved	TiRobot
Schizas et al.18	2012	Prospective	34	92.20%	95.30%	–	No Difference	Mazor
Lonjon et al.19	2016	Prospective	20	92.00%	97.30%	–	Improved	ROSA
Zhang et al.20	2019	Prospective	77	93.60%	98.30%	–	Improved	TiRobot
Kantelhardt et al.21	2011	Retrospective	112	96.50%	98.80%	–	Improved	SpineAssist
Schatlo et al.22	2014	Retrospective	95	87.10%	91.40%	–	No Difference	SpineAssist
Kim et al.23	2015	RCT	40	91.25%	95%	–	Improved	Renaissance
Solomiichuk et al.24	2017	Retrospective	70	83.60%	84.40%	–	No Difference	SpineAssist
Molliqaj et al.25	2017	Retrospective	169	88.90%	93.40%	–	Improved	SpineAssist
Keric et al.26	2017	Retrospective	90	73.50%	90%	–	Improved	Renaissance
Le et al.27	2018	Retrospective	58	86.90%	95.30%	–	Improved	TiRobot
Shillingford et al.28	2018	Retrospective	68	94.90%	97.80%	–	No Difference	Renaissance
Tian et al.29	2017	RCT	40	98.9%	100%	–	Improved	TiRobot
Yang et al.30	2019	Retrospective	60	92.3%	98.5%	–	Improved	Renaissance
Fan et al.31	2020	Prospective RCT	135	91.2%	98.9%	–	Improved	TiRobot
Feng et al.32	2019	Prospective	80	98.5%	98.5%	–	No Difference	TiRobot
Zhang et al.33	2019	Prospective	150	94%	98.4%	–	Improved	TiRobot
Li et al.34	2020	RCT	17	98%	100%	–	No Difference	OrthBot
Chen et al.35	2020	Retrospective	97	92.2%	98.7%	–	Improved	TiRobot
Feng et al.36	2020	RCT	80	93.1%	98.2%	–	Improved	TiRobot
Li et al.37	2022	Retrospective	81	93%	98%	–	Improved	TiRobot
Comparison of pedicle screw accuracy with robotic and navigation guided techniques
Fan et al.38	2018	Retrospective	267	–	91.30%	84.10%	Improved	SpineAssist
Khan et al.39	2019	Retrospective	99	–	99.50%	95.10%	No Difference	Mazor X
Mao et al.40	2020	Retrospective	85	–	97.50%	94.81%	No Difference	Mazor X
Du et al.41	2019	Retrospective	82	–	96.30%	95.80%	No Difference	Mazor X
Comparison of pedicle screw accuracy with free hand, robotic and navigation guided techniques
Roser et al.42	2013	Prospective RCT	112	97.50%	99%	92%	Improved	SpineAssist
Laudato et al.43	2018	Retrospective	84	70.40%	78.80%	69.60%	No Difference	Mazor
Fan et al.44	2017	Retrospective	190	78.03%	94.32%	90.60%	Improved (RG vs FH)	SpineAssist
Comparison of pedicle screw accuracy with free hand and navigation guided techniques
Yang et al.45	2020	Retrospective	72	78.48%	–	97.70%	Improved	O-Arm
Peng et al.46	2020	Retrospective	40	85.20%	–	95.80%	Improved	O-Arm
Jing et al.47	2019	Retrospective	60	90.00%	–	95.81%	Improved	O-Arm
Chen et al.48	2019	Retrospective	45	86.50%	–	96.40%	Improved	O-Arm
Wang et al.49	2019	Retrospective	41	88.70%	–	96.90%	Improved	O-Arm
Zhao et al.50	2018	Retrospective	27	81.58%	–	87.80%	Improved	O-Arm
Urbanski et al.51	2018	Retrospective	49	96.35%	–	95.78%	No Difference	O-Arm
Tajsic et al.52	2018	Retrospective	152	96.62%	–	99.38%	Improved	O-Arm
Knafo et al.53	2018	Retrospective	198	73.80%	–	82.80%	Improved	O-Arm
Xiao et al.54	2017	Retrospective	1208	98.37%	–	95.79%	No Difference	O-Arm
Liu et al.55	2017	Retrospective	138	48.52%	–	72.95%	Improved	O-Arm
Verma et al.56	2016	Retrospective	357	91.21%	–	99.07%	Improved	O-Arm
Ohba et al.57	2016	Retrospective	28	88.88%	–	96.72%	Improved	O-Arm
Liu et al.58	2016	Retrospective	138	82.16%	–	93.02%	Improved	O-Arm
Jin et al.59	2016	Retrospective	32	66.94%	–	79.34%	Improved	O-Arm
Shin et al.60	2015	Prospective RCT	40	92.02%	–	98.38%	Improved	O-Arm
Boon Tow et al.61	2015	Prospective	28	93.40%	–	92.10%	No Difference	O-Arm
Houten et al.62	2012	Retrospective	94	87.23%	–	97.07%	Improved	O-Arm
Shin et al.63	2012	Retrospective	69	94.11%	–	97.16%	No Difference	O-Arm
Ughwanagho et al.64	2012	Retrospective	42	91.00%	–	97.00%	Improved	O-Arm
Silbermann et al.65	2011	Retrospective	67	94.64%	–	98.93%	Improved	O-Arm
Allam et al.66	2013	Retrospective	45	89.80%	–	98%	Improved	3DFL
Amiot et al.67	2000	Retrospective	150	85%	–	95%	Improved	CT
Cui et al.68	2012	Retrospective	49	89.40%	–	94.82%	Improved	CT
Fichtner et al.69	2017	Retrospective	2232	95.62%	–	98.65%	Improved	3DFL
Fu et al.70	2008	Retrospective	24	93.20%	–	96.10%	No Difference	CT
Han et al.71	2010	RCT	42	83.33%	–	96.43%	Improved	CT
Innocenzi et al.72	2017	Retrospective	203	91.81%	–	92.66%	No Difference	CT
Noriega et al.73	2017	RCT	114	89.70%	–	96.40%	Improved	CT
Rajasekaran et al.74	2007	RCT	33	77%	–	98%	Improved	3DFL
Waschke et al.75	2013	Retrospective	1006	86.45%	–	95.15%	Improved	CT
Wu et al.76	2017	Prospective	99	93.33%	–	100%	Improved	CT
Guedes et al.77	2015	Retrospective	80	71.50%	–	98%	Improved	CT
Fraser et al.78	2010	Retrospective	42	73.70%	–	90.90%	Improved	3DFL
Jin et al.79	2016	Retrospective	51	92.35%	–	97.46%	Improved	3DFL
Laine et al.80	2000	RCT	91	86.60%	–	95.40%	Improved	CT

3.2.2. Three dimensional (3D) CT vs FH

The two-dimensional (2D) Fluoro Navigation (FluoroNav) device only showed images in two dimensions, and there were insufficient visuals in different formats suitable for use during surgery to help insert pedicle screws. It appeared that the development of 3D FluoroNav combined the benefits of CT and 2D fluoroscopy-based assistances.⁷⁸ This system provided three-dimensional images and reduced the amount of additional preoperative preparation. Although there were fewer studies on 3D Fluoro Nav than CT Nav, the majority of studies indicated that 3D Fluoro Nav may be superior to the other two guiding systems.^69,74,78,79(Table 1) Similar to CT Nav, 3D Fluoro Nav provided intraoperative three-dimensional images that could be used to guide minimally invasive spinal interventions or assist with pedicle screw insertion into abnormal pedicles. Rajasekaran et al.⁷⁴ conducted an RCT to compare the accuracy of pedicle screw insertion with and without 3D Fluoro Nav in patients with scoliosis or kyphosis. 54 pedicle breaches (23%) occurred in the non-navigation group against just 5 (2%) in the navigation group (P = 0.001). The anterior or lateral cortex had been penetrated by 16% of the screws in the non-navigation group, as well, while only two screws (0.8%) in the navigation group had done so. The 3D Fluoro Nav possessed the advantages of both CT Nav and 2D Fluoro Nav; in theory, it could achieve superior results than other navigation techniques. A significantly large cohort study involving 1084 screw placements using either CT-based navigation (7.34% perforation rate) or 3D fluoroscopy-based navigation (6.6% perforation rate) showed no statistically significant difference in breach rates between the two groups (P = 0.0936).⁷² Both 2D and 3D Fluoro Nav systems were found to be equally safe and accurate in a retrospective analysis, thus the decision between the two may come down to the surgeon's preference and the system's availability in the operating room.⁷⁹ Guedes et al. reported that 3D Fluoro Nav significantly enhanced the accuracy of pedicle screw insertion when compared to conventional, CT Nav, and 2D Fluoro Nav techniques.⁷⁷

3.2.3. Robotic guided (RG) VS navigation (NAV) VS free hand techniques

The potential for robotic assistance to increase implant placement accuracy during spinal surgery has piqued medical professionals' and patients' interests.³⁷ The accuracy of pedicle screw placement with robotic platforms is comparable to and possibly superior to, fluoroscopic-guided or conventional freehand procedures, according to recent literature reviews.14, 15, 16, 17, 18, 19, 20, 21, 22, 23, 24, 25, 26, 27, 28, 29, 30, 31, 32, 33, 34, 35, 36, 37 (Table 1) In the majority of studies, the Gertzbein-Robbins scale (GRS) has been used to evaluate accuracy which measures pedicle/cortical breach based on a perfected and standardized trajectory. As the first robotic system designed for spinal surgery, Mazor robots continue to be the most thoroughly studied to date. Several RCTs have been conducted to confirm the accuracy of these systems. Ringel et al.¹³ published the first RCT, which evaluated 298 pedicle screws implanted in 60 patients using Mazor SpineAssist, in 2012. It had 85% screw placement accuracy, compared to 93% in the freehand group. The robotic group's precision was reduced by the robot's poor fixation and movement of the cannula at the screw insertion site. Thus, mismatched screws deviated laterally. In a similarly constructed trial by Hyun et al.,¹⁴ the Mazor Renaissance robotic model had a 100% accuracy rate (130 pedicle screws put in 30 patients) and the free hand group had 98.6% (140 screws placed in 30 patients). In 2013, Roser et al.⁴² conducted a 3-arm randomized controlled study (RCT) comparing robot-assisted screw placement (n = 18), fluoroscopy-guided (n = 9), and freehand (n = 10) with accuracies of 99.0%, 92.0%, and 97.5%, respectively. A Preliminary study on the newer ROSA robotic system seems promising. Lonjon et al.¹⁹ conducted the only prospective, nonrandomized, case-matched study comparing ROSA (40 screws in 20 patients) and freehand pedicle screw placement (50 screws in 10 patients). The robotic group had a higher rate of accuracy (97.3%) than the freehand group (92.0%), but this difference was not statistically significant (P = 0.639). In this study, the robotic placement of four additional screws was attempted but abandoned due to technical difficulties. Although no prospective, randomized studies of the Excelsius GPS have been reported, initial case reports and limited case series indicate high accuracy and safety with rates comparable to other robotic systems.⁸⁶ In addition, with spinal robotics, the placement accuracy of screws can be evaluated not only with the conventional GRS, but also by measuring deviation from a pre-planned trajectory. Using this novel metric to quantify precision, preliminary results indicate that the Excelsius GPS permits accurate screw placement.⁸⁶

In this narrative review, the authors noted that 49 out of 68 studies showed increased precision with computer-assisted techniques whereas 18 studies showed no difference and only 1 study showed decreased precision, as compared to the freehand technique, summarising an overall advantage of these techniques over freehand pedicle screw placement.

3.3. Radiation exposure

3.3.1. CT vs. freehand

When contemplating the use of a specific intraoperative imaging modality in spinal surgery, radiation exposure to the patient and operating room personnel is a concern. CT Nav requires a preoperative CT scan. Patients' radiation exposure largely hinges on the CT protocol. Solomiichuk et al.²⁴ determined, by analysing the radiation of three various CT protocols-based navigation systems, that the spiral form was suggested for use with CT Nav. Theoretically, Fluoro Nav could reduce radiation exposure time compared to conventional fluoroscopy because such systems do not require repetitive intraoperative C-arm movement. In a prospective study conducted by Fraser and Gebhard et al.,⁷⁸ it was determined that 2D Fluoro Nav emits less radiation than conventional C-arm but more radiation than CT Nav. In addition, 3D Fluoro Nav was observed to have a lower intraoperative radiation dose than other navigation techniques.

The International Commission on Radiological Protection recommends total body exposure of less than 5 Rem per year and extremity exposure of less than 50 Rem per year.⁷⁷ Radiation exposure to the surgeon and OR staff is highly variable and dependent on several factors, including the experience of the surgeon and fluoroscopy technician. Image guidance and robotic systems can substantially reduce the surgeon and staff's radiation exposure. For registration, these systems utilize either preoperative or intraoperative CT. To prevent radiation exposure during intraoperative CT, the OR staff and surgeon may leave the room to a secure distance.¹⁷ CT-based navigation systems (including robotic guidance systems) permit the surgery to proceed with minimal additional intraoperative fluoroscopy and may significantly reduce the cumulative radiation dose surgeons are exposed to. Several previous studies have demonstrated that the use of intraoperative navigation systems reduces radiation exposure to the patient and surgeon (Table 4). Smith et al.⁸⁷ compared radiation exposure in pedicle screw placement using either 2D fluoroscopy or CT navigation and discovered a substantially lower mean radiation exposure to the surgeon's torso in the CT navigation group (0.33 vs. 4.33 mRem) compared to the fluoroscopy group. In 2014, Villard et al.⁸⁸ conducted a prospective, randomized study comparing radiation exposure when using 3D navigation versus 2D fluoroscopic techniques performed manually. In the 2D fluoroscopy group, radiation exposure to the surgeon during pedicle screw implantation was 9.96 times higher than in the navigation group. In addition, the only surgeon radiation exposure in the navigation cohort occurred during the positioning of the interbody device, when a confirmatory fluoroscopy was performed. Bratschitsch et al.⁸⁹ conducted a prospective evaluation of cumulative radiation dose to OR staff using C-arm fluoroscopy versus O-arm navigation in 2019. In comparison to the fluoroscopy cohort, the radiation exposure of surgeons in the navigation cohort was significantly lower. They concluded that, compared to fluoroscopy, a spinal surgeon using navigation could perform up to 10 times as many operations before reaching the maximum permissible annual effective radiation dosage.

Table 4.

Comparison of Radiation exposure time between different techniques used for pedicle screw insertion. FH – Freehand, RG – Robotic guided, NV – Navigation.

Author	Year	Study type	Patients	Radiation exposure (Time/screw)			Significance
				FH	Robotic	Navigation
Comparison of radiation exposure time with free hand and robotic guided techniques
Ringel et al.13	2012	Prospective RCT	60	1.9 min	1.9 min	–	No Difference
Hyun et al.14	2016	Prospective RCT	60	13.3 s	3.5 s	–	Decreased
Wang et al.15	2017	Prospective RCT	30	0.1 min	0.1 min	–	No Difference
Han et al.17	2019	Prospective RCT	234	71.5 s	81.5 s	–	No Difference
Schizas et al.18	2012	Prospective	34	14.2 s	16.7 s	–	No Difference
Lonjon et al.19	2016	Prospective	20	0.4 min	1.23 min	–	Increased
Zhang et al.20	2019	Prospective	77	70.5 s	93.5 s	–	Increased
Kantelhardt et al.21	2011	Retrospective	112	1.28 min	0.56 min	–	Decreased
Solomiichuk et al.24	2017	Retrospective	70	2.1 min	2.3 min	–	No Difference
Keric et al.26	2017	Retrospective	90	56.4 s	24 s	–	Decreased
Le et al.27	2018	Retrospective	58	1.29 min	2.38 min	–	Increased
Alaid et al.81	2018	Retrospective	206	157.13 s (Total)	123.65 s (Total)	–	Decreased
Zhang et al.33	2019	Prospective	150	75.4 s	85.3 s	–	Increased
Feng et al.32	2019	Prospective	80	47.45 s	38.87 s	–	Decreased
Li et al.34	2020	RCT	17	1.04 s	0.56 s	–	Decreased
Jamshidi82	2021	Retrospective	485	151.9 s	57.3 s	–	Decreased
Li et al.37	2022	Retrospective	81	0.43 min	0.37 min	–	Decreased
Comparison of radiation exposure time with free hand, robotic and navigation guided techniques
Roser et al.42	2013	Prospective RCT	112	31.5 s	15.98 s	10.36	Decreased
Fan et al.44	2017	Retrospective	190	8.89 min	4.02 min	6.36	Decreased

3.4. RG vs REST

In spinal surgery, fluoroscopy is essential for the localization and guidance of instruments. Consequently, patients, surgeons, and staff in the operating room during spinal surgery are substantially more exposed to harmful ionizing radiation than during other neurosurgical subspecialties or orthopaedic surgery. Although the patient still receives radiation exposure as part of the procedure, albeit minimal, with the invention of O-arm, robotic spine surgery can minimize or even eliminate radiation exposure for the surgeon and operating room staff during the procedure. Numerous studies have evaluated the radiation exposure associated with robot-assisted screw instrumentation; however, radiation exposure appears to differ between studies and platforms.13, 14, 15^,17, 18, 19, 20, 21 (Table 4), Keric et al.²⁴ discovered that the use of robotic assistance could substantially reduce intraoperative radiation time (mean difference 12.38, 95% CI 17.95 to 6.80; P 0.0001) and intraoperative radiation dosage (standard mean difference 0.64, 95% CI 0.85 to 0.50; P 0.0001). Importantly, as the number of cases treated with robot assistance increases, the average radiation time spent per patient decreases, suggesting a learning curve effect and the possibility of significantly minimizing radiation hazards as surgeons gain experience with these novel approaches.²⁷

Comparing robotic surgery to manual or fluoroscopic surgery has yielded similar results in previous research. In 112 patients, Kantelhardt et al.²¹ conducted a retrospective analysis of conventional open-to-open RG and percutaneous RG pedicle screw placement. The authors reported that RG reduced radiation exposure and improved screw positioning precision. In a brief prospective cohort study involving 37 patients, Roser et al.⁴² published a three-arm randomized trial comparing freehand, navigation, and robotic-assisted procedures. While there were diminished fluoroscopy time and radiation dosages in the navigation and robotic groups, the authors found no significant differences in precision. Similar findings have also been published in the literature on orthopaedic trauma. Wang et al.¹⁵ conducted a prospective randomized comparison of robotic assisted sacroiliac screw fixation and conventional freehand sacroiliac screw fixation. In comparison to the freehand group, the robotic-assisted cohort demonstrated increased precision and decreased radiation exposure durations.

The authors of the current narrative review conclude that 10 out of 19 studies show decreased radiation exposure with the help of computer-assisted surgeries, thus highlighting its advantage over the freehand technique. However, with better training and experience in handling computer-assisted surgeries over time, radiation exposure can be decreased by manifolds.

3.5. Blood loss and operative time

Several studies have attempted to quantify and compare intra-operative blood loss between robotic and navigation-guided techniques and free-hand techniques (Table 3), but only Fan et al.⁴⁴ has compared all three arms in a single study. They found that robotic-guided surgical techniques had statistically significantly reduced blood loss as compared to the other two arms. On the contrary, there are a few studies that report increased blood loss with CT-guided navigation systems as compared to free hand techniques and this was attributed to lengthy working times.^{76,78,80,83,85} In this review, 11 of 23 studies mentioned decreased blood loss with computer-assisted surgeries, whereas 5 studies showed no significant difference.

Table 3.

Comparison of Intra-operative blood loss between various procedures. FH – Freehand, RG – Robotic guided, NV – Navigation.

Author	Year	Study type	Patients	Blood loss (ml, average)			Significance
				FH	RG	NV
Comparison of intra-operative blood loss with free hand and robotic guided techniques
Zhang et al.20	2019	Prospective	77	362	171.6	–	Decreased
Feng et al.32	2019	Prospective	80	356.25	171.6	–	Decreased
Li et al.34	2020	RCT	17	245	257	–	No Difference
Chen et al.35	2020	Retrospective	97	573	499	–	Decreased
Feng et al.36	2020	RCT	80	237.5	165	–	Decreased
Fan et al.31	2020	Prospective RCT	135	210	220	–	No Difference
Li et al.37	2022	Retrospective	81	273.41	248.65	–	Decreased
Comparison of intra-operative blood loss with robotic and navigation guided techniques
Fan et al.38	2018	Retrospective	267	–	681	669	No Difference
Khan et al.39	2019	Retrospective	99	–	85.2	100.3	Increased
Mao et al.40	2020	Retrospective	85	–	1098	1380	Increased
Comparison of intra-operative blood loss with free hand, robotic and navigation guided techniques
Fan et al.44	2017	Retrospective	190	557	362	528	Decreased
Comparison of intra-operative blood loss with free hand and navigation guided techniques
Yang et al.45	2020	Retrospective	72	350	–	90.7	Decreased
Peng et al.46	2020	Retrospective	40	255.91	–	88.33	Decreased
Chen et al.48	2019	Retrospective	45	186.67	–	109.76	Decreased
Wang et al.49	2019	Retrospective	41	838.1	–	822.5	No Difference
Miller et al.83	2017	Retrospective	44	389.8	–	790	Increased
Liu et al.55	2017	Retrospective	138	1228	–	1024	Decreased
Rajasekaran et al.74	2007	RCT	33	950	–	1100	No Difference
Wang et al.84	2016	RCT	40	122.5	–	220.5	Increased
Wu et al.76	2017	Prospective	47	390.91	–	415.79	Increased
Fraser et al.78	2010	Retrospective	42	133	–	257	Increased
Khanna et al.85	2016	Retrospective	56	229	–	154	Decreased
Laine et al.80	2000	RCT	91	1107	–	1270	Increased

There are specific supplementary procedures involved when using a surgical robotic system, including the mounting of a patient reference array, synchronization using intra-operative imaging, installing the robotic arm, and registering the tools. Kantelhardt et al.²¹ reported a mean average placement time per screw of 57 min for percutaneous RA screws, 65.2 min for open RG screws, and 52.9 min for FH screws. Lonjon et al.¹⁹ reported that RG screw placement required an additional hour of surgical time compared to FG-screw placement. In contrast, Hyun et al.¹⁴ found no difference between the operative durations for percutaneous RG screw placement and open FG screw placement. The time saved by avoiding surgical exposure may have countered the time required to set up the surgical robot system. The majority of these studies report ‘skin-to-skin’ time for the entire spine surgery, which includes sequential rod placement, decompression, cage implantation, and wound closure (Table 2). This may introduce confounding variables; a superior study design would compare the time between the initial skin incision and the placement of the final screw.

Table 2.

Comparison of Operative time during different types of procedures. FH – Freehand, RG – Robotic guided, NV – Navigation.

Author	Year	Study type	Patients	Operative Time (minutes, average)			Significance
				FH	RG	NV
Comparison of operative time with free hand and robotic guided techniques
Ringel et al.13	2012	Prospective RCT	60	132	151	–	No Difference
Hyun et al.14	2016	Prospective RCT	60	208.5	208.5	–	No Difference
Wang et al.15	2017	Prospective RCT	30	104	150	–	No Difference
Han et al.17	2019	Prospective RCT	234	138	149.5	–	No Difference
Lonjon et al.19	2016	Prospective	20	112	186	–	Increased
Zhang et al.20	2019	Prospective	77	154.7	165.3	–	No Difference
Kantelhardt et al.21	2011	Retrospective	112	52.9 s/screw	59.1 s/screw	–	No Difference
Schatlo et al.22	2014	Retrospective	95	189	205	–	No Difference
Kim et al.23	2015	RCT	40	195	217.5	–	Increased
Solomiichuk et al.24	2017	Retrospective	70	264.1	226.1	–	No Difference
Keric et al.26	2017	Retrospective	90	218.87	202.55	–	No Difference
Le et al.27	2018	Retrospective	58	119.5	199.1	–	Increased
Alaid et al.81	2018	Retrospective	206	219	184	–	No Difference
Jamshidi et al.82	2021	Retrospective	485	98.8	118.3	–	No Difference
Tian et al.29	2017	RCT	40	118.2	138.9	–	No Difference
Fan et al.31	2020	Prospective RCT	135	210	220	–	No Difference
Feng et al.32	2019	Prospective	80	117.6	111.6	–	No Difference
Zhang et al.33	2019	Prospective	150	117.8	184.7	–	Increased
Li et al.34	2020	RCT	17	266	289	–	No Difference
Chen et al.35	2020	Retrospective	97	291.9	283.1	–	Decreased
Feng et al.36	2020	RCT	80	230.63	196.25	–	Decreased
Li et al.	2022	Retrospective	81	7.25/Screw	5.58/Screw		Decreased
Comparison of operative time with robotic and navigation guided techniques
Fan et al.38	2018	Retrospective	267	–	239	228	Decreased
Khan et al.39	2019	Retrospective	99	–	153.9	162.1	Increased
Mao et al.40	2020	Retrospective	85	–	377	329	Decreased
Comparison of operative time with free hand, robotic and navigation guided techniques
Roser et al.42	2013	Prospective RCT	112	111.2	140.8	160.8	Increased
Fan et al.44	2017	Retrospective	190	178	201	194	Increased
Comparison of operative time with free hand and navigation guided techniques
Yang et al.45	2020	Retrospective	72	120.6	–	134.3	Increased
Peng et al.46	2020	Retrospective	40	132.27	–	201.67	Increased
Chen et al.48	2019	Retrospective	45	129.79	–	217.62	Increased
Wang et al.49	2019	Retrospective	41	255.19	–	222.55	Decreased
Tajsic et al.52	2018	Retrospective	152	193	–	218	Increased
Knafo et al.53	2018	Retrospective	198	66.1 s/level	–	57.3 s/level	Decreased
Miller et al.83	2017	Retrospective	44	193.98	–	200.79	Increased
Xiao et al.54	2017	Retrospective	1208	4.4 h	–	4.3 h	No Difference
Liu et al.55	2017	Retrospective	138	243	–	265	Increased
Jin et al.59	2016	Retrospective	32	257	–	268	Increased
Silbermann et al.65	2011	Retrospective	67	193	–	183	Decreased
Rajasekaran et al.74	2007	RCT	33	4.61/screw	–	2.37/screw	Decreased
Wang et al.84	2016	RCT	40	134.4	–	124.5	Decreased
Wu et al.76	2017	Prospective	47	247.55	–	294.68	Increased
Guedes et al.77	2015	Retrospective	80	270	–	320	Increased
Fraser et al.78	2010	Retrospective	42	231	–	321	Increased
Jin et al.79	2016	Retrospective	51	264	–	234	Decreased
Khanna et al.85	2016	Retrospective	56	199	–	177	Decreased
Laine et al.80	2000	RCT	91	179	–	160	Decreased

Our review found that only 13 studies noted decreased operative time, 16 studies showed no difference, whereas 17 studies showed increased operative time. Thus, computer-assisted surgeries may not have the advantage of reducing the operative time, specifically, as compared to the freehand technique.

3.6. Length of hospital stay

One of the major advantages of robotic spine surgery is that it reduces the length of stay in the hospital.^31,34,35 This is because it is minimally invasive, which means that it requires smaller incisions and causes less damage to the surrounding tissues. As a result, patients who undergo robotic spine surgery experience less pain and discomfort, and they can recover more quickly.⁸² Additionally, because robotic spine surgery is more precise and accurate than free-hand techniques, there is less of a chance for complications or errors during the surgery. This helps to further reduce the patient's hospital stay, as it can minimize the need for additional procedures or follow-up visits.²⁰ The average time to ambulate after freehand surgeries was found by Kim et al. to be 39.7 h, while it was 36.2 h after RG surgery (p = 0.363). FH operations have been shown to result in a higher length of stay (9.4 days vs. 6.8 days, p = 0.020), as reported by Hyun et al.¹⁴

According to Kantelhardt et al.²¹ the length of time it took for patients to be discharged from the hospital following surgery was 14.6 days for those who underwent open surgery and collectively received 286 screws, 11.6 days for those who underwent robot-guided open surgery and collectively received 94 screws, and 10.1 days for those who underwent robot-guided percutaneous operations and collectively received 156 screws, with a p-value of 0.009 for the difference. However, various studies report that there is no statistically significant difference when these two techniques are compared in terms of duration of stay in the hospital.^17,31,34,82 (Table 5).

Table 5.

Comparison of Length of stay between various procedures. FH – Freehand, RG – Robotic guided, NV – Navigation.

Author	Year	Study type	Patients	Length of stay (Days, Average)			Significance
				FH	RG	NV
Comparison of length of stay with free hand, robotic guided techniques
Hyun et al.14	2016	Prospective RCT	60	9.4	6.8	–	Decreased
Han et al.17	2019	Prospective RCT	234	4.95	4.82	–	No Difference
Zhang et al.20	2019	Prospective	77	5.6	5.1	–	Decreased
Jamshidi et al.82	2020	Retrospective	485	2.6	2.7	–	No Difference
Chen et al.35	2020	Retrospective	97	13.7	12.8	–	Decreased
Fan et al.31	2020	Prospective RCT	135	6	5	–	No Difference
Li et al.34	2020	RCT	17	12.8	13.1	–	No Difference
Comparison of length of stay with free hand, robotic and navigation guided techniques
Fan et al.44	2017	Retrospective	190	8.9 + 1.8	6.3 + 1.2	7.9 + 1.1	Decreased (FH vs RG)

The authors of this present narrative review noted that only 4 of 8 studies showed reduced length of stay in the hospital, whereas the remaining studies showed no significant difference. Thus, computer-assisted surgeries may or may not affect the hospital stay as compared to freehand techniques.

3.7. Functional outcomes and complications

Very few studies have compared the functional outcomes following surgeries performed by different techniques.^85,90,91 Park et al.⁹⁰ quoted that at the final follow-up, both groups (Robotic guided and Freehand technique) reported significant reductions in their VAS scores for both the back and the leg pain (both p < 0.001), although there were no significant differences between the groups in terms of back pain (VAS-p = 0.876) or leg pain (VAS-p = 0.429). Even while there was a statistically significant increase in ODI scores, by the end of the follow-up period (p < 0.001), there was no difference between the groups overall. (p = 0.952) Significant improvements were seen over time for the back VAS, leg VAS, and ODI. (p < 0.001, for every single variable at each follow-up compared to baseline).

In their study, Kantelhardt et al.²¹ found that post-operative infections occurred in 10.7% of open non-robotic procedures compared to 2.7% of robot-guided open procedures. This difference was statistically significant (p = 0.047). The need for revision operative intervention was only necessary in 10 of these cases (0.6% of the percutaneous procedures, 12.6% of the open robotic-guided operations, and 12.2% of the open procedures).

Data on infections was also released by Keric et al.²⁶ Their study showed that 10.6% of patients in the robot-assisted percutaneous group had a surgical site infection, compared to 20.8% of patients in the open procedures group (p = 0.104). Only antibiotics were necessary to treat only one of the open procedures, whereas surgical intervention was required in four of these cases. Patients undergoing robot-assisted surgery were required to undergo revision surgery in four of the cases, whereas antibiotics were sufficient to treat three of the complications that arose from these procedures. The bulk of the difficulties occurred as a result of technical challenges with the robot (failure of hardware or software, inability to successfully register fluoroscopy and CT), or cannula sliding, which led to screws being placed in the incorrect location.29, 30, 31, 32, 33 Clinical problems included hemothorax, pulmonary embolism, and leakage of cerebral spinal fluid (CSF).^34,35 Devito et al.⁹² found that the use of SpineAssist (RG) for pedicle screw instrumentation resulted in reversible neurological problems in just four out of 593 patients (0.7% of total cases). When weighed against robot groups, Freehand operations showed a greater rate of Dural tears (four compared to one) across the course of 25 procedures (p = 0.142).

In a fluoroscopy-guided procedure, Schatlo et al.²² and Solomiichuk et al.²⁴ each reported one incident of nerve root injury; however, in the robot-assisted cohort, there was not a single instance of this complication.

In recent years, there has been growing concern regarding the potential of adjacent segment disease (ASD) associated with facet joint violation.⁹⁰ Robot-guided spine surgery was related to fewer proximal facet joint violations, and it allowed for a larger facet-to-screw distance. This was in comparison to the freehand procedure. According to the research carried out by Zhang et al.,³³ of those in the free-hand group experienced much more facet joint violations than those in the robot-guided group. Additionally, the average distance of pedicle screws from facet joints was greater in the robot-guided group than it was in the free-hand group, measuring 4.16 mm as opposed to 1.92 mm. Similar outcomes were reported by Kim et al.¹⁶ The results of this study showed that neither the robot-assisted nor the conventional procedures provided significantly different ASD outcomes, and the difference in revision surgery between the two groups (0 patients vs. 2 patients) was statistically insignificant.

This current extensive literature review found that the use of Computer-assisted surgeries can cause a significant reduction in rates of infections, neurological deficits caused by pedicle screw violations, the need for revision surgeries, and rates of radiological ASD, as compared to Freehand techniques.

3.8. Learning curve

Experience and familiarity with the robot have been found to positively correlate with screw placement precision. In a comparable study designed to assess the learning curve related to computer-assisted navigation, the curve declined substantially after 6 months and leveled off after 12 months for both a 10-year and 20-year spine surgeon's level of expertise.⁷³ One surgeon's lumbar pedicle screw cortical perforation rate decreased by 3.9% (P = 0.006) and the other surgeon's rate decreased by 5.6% (P < 0.001).⁷⁵ Similarly, the operative time was reduced by 20.9 min and 40.3 min (P < 0.001) for the computer-assisted navigation groups respectively. Separately, it has been demonstrated that skilled robot control can contribute to a reduction in radiation exposure time.⁸¹ Kim et al.¹⁶ determined that the total fluoroscopy duration decreased by 30% after the initial eight cases. For this reason, they suggested that a minimum of ten cases are required to acquire the necessary experience to minimize radiation exposure time.

Any new technology necessitates familiarization with its application by the surgeon before it can produce consistent results. In their subsequent cases, Devito et al.⁹² reported an improvement in screw accuracy, execution rate, and operative time. Hu et al.⁹³ divided the 150 patients who underwent the insertion of a RG pedicle screw by a single surgeon into five groups of 30 patients each. The outcome measures included frequency of incorrect placement, rate of successfully inserted screws, and rate of exchange to manual placement. Following the initial 30 procedures, the rate of successfully inserted screws increased and the rate of conversion to manual placement decreased with expertise. Schatlo et al.²² reported that after the surgeons completed their 25th case, the rate of misplaced screws with RG decreased considerably. The authors suggested that the first 25 cases of robotic spine surgery conducted by a new surgeon should be supervised by an expert.

Considering the above data, the authors of this review recommend conducting a sizeable number of computer-assisted surgeries as the learning curve may not be as long.

3.9. Cost benefit analysis

The contribution that any new technology will make to the existing healthcare system will be determined by the complicated interaction of a number of factors, including the cost, accessibility, and quality of the technology.⁸⁴ It is difficult to evaluate the true contribution that robotic spine surgery actually makes to the healthcare system since there is a dearth of accurate data on the long-term outcomes of robotic spine surgery, particularly in terms of how it compares to conventional approaches. Opponents of robotic spine surgery argue that the high initial cost of a surgical robot system is not outweighed by the savings in healthcare expenditures that occur from a lower infection rate, shorter hospital stays, and a decreased need for revision surgery to fix misplaced screws.⁸⁵ However, these savings are said to be more than the high initial cost of a surgical robot system. In addition, the purchase of cutting-edge technology can give hospitals the ability to establish robotics programs and market themselves as providers of cutting-edge spine surgery procedures. According to a cost-effectiveness model, robotic spine surgery resulted in total savings of $608,546 over the course of a year in an active neurosurgery practice that handled 557 patients annually.⁹⁴ On the other hand, the assumptions made by the authors regarding how to translate a number of clinical outcome metrics into cost reductions cannot be transferred to other healthcare settings due to the fact that there are regional and cultural differences between the settings.⁴³

4. Future implications

Due to its growing use and development, robotic spine surgery holds promise. Robots are being used for spinal tumor resections, radiofrequency ablations, and deformity surgeries.⁶⁶ To make robots more accessible, manufacturers are making them lighter, portable, and cheaper.⁷² The Internet of Skills and Artificial Intelligence (AI) may be used in robotics' next era. The Internet of Skills allows remote surgical treatments with latencies of less than 10 ms at distances up to 1500 km using ultra-fast, low-latency 5G connectivity.⁹¹ However, AI and machine learning (ML) can educate robots to predict surgical motions, allowing surgeons to operate from longer distances with less delay. This technology can also be used in surgical training programs to let trainees watch online surgeries as if they were onsite. In this context, several virtual reality (VR) and augmented reality (AR) tools have been introduced into surgical training programs, and some of these simulation technologies have been connected to improvements in trainees' operating times and performance.⁹⁴

Madhavan et al.⁹⁵ proposed a VR and AR-enabled semi-independent spine robot for neurosurgery. While watching the surgical procedure using an augmented reality headset or vocal instructions, the surgeon may operate the robot. AR headgear can do pre- and intraoperative imaging scans and three-dimensional screw trajectory calculations. The headgear projected target structures onto the patient by combining these two modalities. Ueda et al.⁹⁶ created a robotic simulator for surgeon collision avoidance in Japan. Neurosurgeons and orthopaedic surgeons use thin, long dissection instruments in spinal surgeries, making autonomous collision avoidance useful. The Smart Tissue Autonomous Robot (STAR), is nearly ready for human testing. This robot has demonstrated the capacity to execute autonomously certain surgical procedures on pigs, such as tissue suturing.⁹⁷

As robotics integrates AI, AR, and ML, surgical robot platforms may be able to foresee intraoperative changes, boosting precision and efficacy.⁷⁹ Technology may allow computer-aided pedicle screw placement that considers the patient's anatomy, biomechanics, and surgery results. Open and percutaneous placement of thoracic and lumbar pedicle screws will enable more sophisticated applications such as C1–C2 posterior fusion and S2-pelvic screw implantation.⁸⁰ Future robotics may aid laminectomies, osteotomies, revision, and deformity procedures, as well as extra- or intradural tumor removal.⁹⁶

4.1. Study limitations

This review has the limitations as that of a narrative review. It was also limited to original clinical research available in the English language. To improve the data quality we excluded conference proceedings, reviews, proposed research or protocol description, thesis, abstracts, lectures, commentaries, and editorials from this review.

5. Conclusion

The increased precision afforded by robotic-assisted spine surgery leads to improved patient outcomes. Robotic surgery reduces blood loss and radiation exposure, along with decrease in the rate of complications such as infections, neurological injury, need for revision surgeries and radiological ASD. However only short-term research is available as of now. Certain parameters, such as intraoperative surgical duration and length of hospital stay do not have sufficient studies to prove significant reductions to be deemed advantageous. There is a need for long term follow-up studies so as to better assess clinical results and cost-effectiveness.

Ethical approval

The Authors did not seek any institutional ethical approval, since it is a narrative review.

Use of any AI tool: The Authors refrained from using any AI tools for writing the manuscript.

Authors’ contribution

Dr. H S Chhabra provided the detailed structure and concept of the article with meticulous proof-reading and suggestions, Dr Amogh Zawar provided the writing assistance and literature search, Dr Anuj Mundra, Dr Sachin Sharma and Dr Kalyan kumar Varma provided the proof-reading of the article.

Funding

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Declaration of competing interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.