This systematic review was conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines. The protocol was registered on the International Prospective Register of Systematic Reviews (PROSPERO) with registration number (CRD420251066149).

The primary outcome was the accuracy of surgical planning and execution, assessed through measures such as anatomical localization and deviation errors. Secondary outcomes included surgical plan modifications, time efficiency (e.g., operative or planning time), radiation exposure, clinical outcomes (e.g., complication rates, blood loss), and user-related outcomes such as surgeon satisfaction and anatomical understanding. Outcomes were analyzed by technology type and surgical application phase (preoperative or intraoperative).

Eligibility criteria were defined according to the PICOS (Population, Intervention, Comparator, Outcome, Study design) framework (14). Population (P) included patients, surgeons, or clinical teams involved in surgical planning. Intervention/Exposure (I) was the use of immersive or interactive technologies (virtual reality, augmented reality, mixed reality, or multi-modal platforms) applied to surgical planning. Comparators (C), where applicable, included conventional planning methods such as two-dimensional imaging, three-dimensional reconstructions, or standard visualization tools. Outcomes (O) included planning accuracy, spatial understanding, surgical decision-making, workflow efficiency, operative metrics, and user-centered outcomes. Study designs (S) included randomized controlled trials, prospective observational studies, feasibility studies, and retrospective analyses.

Studies were eligible for inclusion if they investigated or reported on the use of immersive technologies in any surgical specialty, including observational, feasibility, and comparative research. Studies were excluded if they were non-English publications or did not involve human surgical procedures, focused solely on educational or simulation purposes without clinical application, lacked outcome data, or were non-original publications such as reviews, editorials, conference abstracts, and technical notes.

A comprehensive and systematic literature search was conducted to identify studies that evaluate the effectiveness of immersive technologies for surgical planning before the operation. The databases searched included PubMed, Google Scholar, Web of Science and Ovid MEDLINE from inception until February 20, 2025. The search terms employed a combination of Medical Subject Headings (MeSH), and keywords related to (“Mixed reality” OR “Augmented reality” OR “Virtual reality”) AND (“Surgery” OR “Preoperative planning”). The electronic search strategy is provided in the (Supplementary Appendix A1). Titles and abstracts were independently screened by two reviewers, followed by full-text assessment by the same reviewers. Any disagreements were resolved through discussion and consensus, with consultation of a third reviewer when required. Formal inter-rater agreement statistics were not calculated.

The screening process has now been described in greater detail. Two independent reviewers screened titles and abstracts and subsequently assessed full-text articles for eligibility. Disagreements were resolved through discussion and consensus, with involvement of a third reviewer when necessary.

Formal inter-rater agreement statistics (e.g., Cohen's kappa) were not calculated, as this was not predefined in the study protocol and discrepancies were infrequent and resolved through consensus. This has now been explicitly stated in the Methods section for transparency. Embase, Scopus, and the Cochrane Library were not searched due to substantial overlap with MEDLINE- and Web of Science–indexed journals and institutional access limitations. Given the rapid technological evolution of immersive surgical planning, the selected databases were considered sufficient to capture the relevant peer-reviewed clinical literature.

Two independent reviewers extracted data from each included study using a standardized data collection form. Extracted information included study characteristics (author, year, country, sample size), type of immersive technology used (virtual reality, augmented reality, or mixed reality), imaging modality, 3D software platform, visualization hardware, surgical specialty, and phase of application (preoperative or intraoperative). Outcome data were also collected, including accuracy metrics, surgical plan changes, time efficiency measures, radiation exposure, clinical outcomes, and user-reported experiences.

To evaluate the methodological quality and risk of bias of the included studies, two validated tools were employed based on the study design. RoB 2.0 Tool (Revised Cochrane risk-of-bias tool for randomized trials) was used for assessing randomized controlled trials (RCTs). This tool evaluates five domains: bias arising from the randomization process, deviations from intended interventions, missing outcome data, measurement of the outcome, and selection of the reported result. Each domain is rated as having low, some concerns, or high risk of bias (15).

ROBINS-I Tool (Risk Of Bias In Non-randomized Studies of Interventions) was used to assess prospective observational and feasibility studies. This tool examines seven domains of bias: due to confounding, selection of participants, classification of interventions, deviations from intended interventions, missing data, measurement of outcomes, and selection of reported results. It is suitable for evaluating non-randomized studies that aim to estimate the effects of interventions, providing an overall risk of bias rating as low, moderate, serious, or critical (16). All included studies were independently assessed by two reviewers using the appropriate tool for their design. Any disagreements were resolved through discussion to ensure consistent and accurate judgments.

Extracted data were synthesized narratively due to the heterogeneity of study designs, technologies, procedures, and outcome measures. Studies were grouped based on the type of immersive technology used (virtual, augmented, or mixed reality) and the phase of surgical application (preoperative or intraoperative). Key outcomes such as accuracy, surgical plan modification, time efficiency, radiation exposure, clinical results, and user feedback were summarized and compared across studies. Quantitative findings were tabulated to highlight trends, but no meta-analysis was performed due to variability in outcome reporting and measurement methods.

The initial search across four electronic databases (PubMed, Google Scholar, Web of Science and Ovid MEDLINE) yielded a total of 2,374 records (988 from MEDLINE Ovid, 574 from Web of Science, 568 from PubMed, and 244 from Google Scholar) (Figure 1). After removal of 406 duplicate records, 1,968 studies remained for title and abstract screening. Of these, 1,929 studies were excluded for the following reasons: non-related studies (n = 1,056), systematic reviews or meta-analyses (n = 363), cadaveric studies (n = 237), animal studies (n = 142), and cross-sectional studies (n = 131).

A total of 39 studies were sought for full-text retrieval. Through manual searching, we identified an additional seven studies, bringing the total number of articles subjected to full-text screening to 46. Manual searching included two strategies: First one is backward citation searching of relevant systematic reviews to identify original studies that might have been missed during database searches. While the second one is snowballing (pearl growing), in which the “related articles” and “similar studies” functions in PubMed, Scopus, and Google Scholar were used when retrieving known eligible articles.

Following full-text review, 16 studies were excluded (educational-only interventions without real patients, n = 7 (4, 17–22); unavailable full text, n = 3 (23–25); interventions designed solely for distraction or anxiolysis, n = 2 (26, 27); no relevant outcomes, n = 2 (28, 29); unpublished, n = 1 (30); review article, n = 1 (31)). Ultimately, 30 studies met the inclusion criteria and were included in this systematic review (7, 32–60).

A total of 30 studies were included in this systematic review, encompassing 1,270 participants, including both patients and clinicians (Table 1). The most common study design was the randomized controlled trial, accounting for 14 studies. This was followed by 9 prospective observational studies, 3 prospective feasibility studies, 3 retrospective studies (including case series and observational), and 1 preclinical comparative study. Participants were recruited across a wide geographic distribution, with studies conducted in China, USA, Germany, France, Netherlands, and other countries.

Virtual Reality was the most commonly used technology, applied in 17 studies. Augmented Reality was used in 8 studies, Mixed Reality in 2 studies. Multi-modalities, and other immersive tools such as Extended Reality and FLER (Fluorescence-based enhanced real) combined with NIR (Near-infrared) cameras appeared in isolated cases.

The included procedures covered a diverse range of surgical specialties. Commonly studied interventions included neurosurgery, breast tumor localization, robotic-assisted urological surgeries (e.g., RAPN and RALP), spinal surgery, cholecystectomy, and vascular access techniques. Other notable areas included colorectal, hepatic, orthopedic, thoracic, and cardiovascular surgeries.

Across the 30 included studies, a range of immersive and interactive technologies were employed, primarily VR, AR, and MR, often integrated with preoperative imaging modalities such as CT, MRI, ultrasound, and angiographic datasets (Table 2). Most studies leveraged these technologies for preoperative planning (n = 23), with several extending applications to intraoperative navigation or guidance (n = 14), and a few specifically targeting tumor localization or vascular access. Intraoperative applications often relied on real-time overlays and holographic projections, while preoperative uses focused on anatomical visualization, surgical rehearsal, and spatial orientation.

A variety of 3D software platforms supported model reconstruction and visualization. Frequently used platforms included 3D Slicer, Materialise Mimics, Reveal (Ceevra), OsiriX, and vendor-specific programs such as Pulmo3D, Synapse 3D, and BananaVision. In studies using AR or MR, Microsoft HoloLens was the most common hardware, while VR studies typically employed head-mounted displays (HMDs) such as the Oculus Rift, HTC Vive, and SpectoVR.

Notably, the immersive technologies were tailored to the specific surgical domain and task. Neurosurgical and thoracic procedures often utilized MR and AR for real-time intraoperative support, while urologic, hepatobiliary, and orthopedic applications favored VR-based planning.

Among the 16 RCTs assessed using RoB 2, 10 studies raised to have some concerns, mainly due to the first (D1) and second (D2) domains. Four studies were judged to have an overall low risk of bias, while only 2 studies had an overall high risk of bias, primarily due to the first and second domains once again. No study was rated as high risk overall. The domains of missing outcome data (D3) and reported results selection (D4) were generally well-addressed across trials.

The risk of bias across the remaining 14 included studies was assessed using the ROBINS-I tool. Most studies exhibited significant methodological limitations, with 9 of the 14 studies rated as having a serious overall risk of bias and 4 studies judged to have critical risk of bias. Critical and serious risks were most frequently observed in the “bias due to confounding” domain. Bias in the selection of participants was primarily rated as moderate (9 studies), while the third (D3), fourth (D4), and fifth (D5) domains were predominantly judged as low risk for almost all of the studies.

Outcomes related to spatial understanding and preoperative planning accuracy were most frequently evaluated in randomized controlled trials, whereas workflow efficiency, feasibility, and user experience outcomes were predominantly reported in non-randomized studies. These differences were considered when interpreting the strength of evidence across outcomes. Finally, while the randomized studies demonstrated generally solid internal validity, the non-randomized evidence carried higher susceptibility to bias. These assessments highlight the need for cautious interpretation of observational findings and support the strength of conclusions drawn primarily from the randomized trials (Supplementary Appendix A2).

Out of the 30 included studies, 17 [9 RCTs (53%) and 8 observational/feasibility studies (47%)] reported quantitative or qualitative outcomes related to accuracy in surgical planning and execution as demonstrated in Table 3A. A wide range of procedures were evaluated, including cranial, spinal, maxillofacial, breast, hepatobiliary, thoracic, and orthopedic surgeries.

Augmented Reality demonstrated consistent improvements in surgical accuracy across several domains. Notable findings include superior osteotomy positioning in mandibular distraction (p < 0.01), enhanced perforator identification in breast reconstruction (62% vs. 41%, p = 0.02), and improved pedicle screw placement accuracy in spine surgery, with 98% of screws rated as “excellent” compared to 91.7% in the control group (p < 0.0003). AR was also shown to be equally effective as ultrasound for breast tumor localization, and it enabled accurate preoperative planning in microsurgical flap procedures and spinal instrumentation.

Virtual Reality was similarly effective in enhancing spatial understanding and reducing deviation. For instance, VR significantly reduced deviation in mandibular condylar fracture fixation (0.639 mm vs. 0.995 mm, p < 0.001), improved quadrant accuracy for breast tumor localization in experienced surgeons (p = 0.01), and achieved 100% intraoperative correlation in thoracoscopic segmentectomy. In robotic cholecystectomy, VR facilitated improved biliary anatomy identification, recognizing anatomical variants with a high confirmation rate (p = 0.0037), and it improved MRCP interpretation in another cholecystectomy study (p < 0.05).

Mixed Reality also showed benefits, particularly in reducing needle adjustments during thoracic epidural placement (7.2 vs. 14.4 movements, p = 0.01). One study using FLER found significant correlation with perfusion markers, enabling better intraoperative ischemia detection (p < 0.05).

Across the 17 studies reporting accuracy outcomes, 16 (94%) demonstrated a favorable directional effect for immersive technologies compared with conventional approaches. Quantitative improvements included reduced deviation, higher placement grades, and improved localization success, with reported absolute differences ranging from 6% to 21% depending on procedure. No study reported inferior accuracy with immersive technology, supporting a consistent directional benefit despite heterogeneity in outcome metrics.

A total of 16 [8 RCTs (50%) and 8 observational/feasibility studies (50%)] studies reported on the impact of immersive technologies on time efficiency, primarily focusing on reduced operative times, faster imaging interpretation, and improved procedural workflows (Table 3B). These studies involved various surgical specialties, including hepatobiliary, orthopedic, neurosurgical, vascular, and reconstructive procedures.

Augmented Reality was associated with consistent time-saving benefits in intraoperative and preoperative contexts. For example, AR significantly reduced flap harvest time in breast reconstruction (136 ± 7 vs. 155 ± 7 min, p = 0.012) and mapping time (2.3 vs. 20 min, p < 0.001). In spinal surgery, AR decreased pedicle screw placement time by nearly 50% (p < 0.05), and in ultrasound-guided vascular access, AR reduced the completion time from 18.5 to 11.5 s (p = 0.04). Similarly, vertebroplasty procedures showed reduced fluoroscopy time with AR (p = 0.005).

Virtual Reality contributed to streamlined planning and faster execution in several procedures. VR reduced operative time in mandibular fracture fixation (180.6 vs. 223.2 min), MRI interpretation time in breast tumor localization (p < 0.001), and operative time in tibial plateau fracture management (156 vs. 172 min, p < 0.001). However, some studies reported non-significant time gains, such as clamp time reduction in robotic-assisted nephrectomy.

Mixed Reality also demonstrated efficiency gains. In thoracic epidural placement, MR shortened the procedure duration (4.5 vs. 7.3 min, p = 0.02). Additionally, technologies like FLER offered rapid perfusion analysis with an average processing time of 7.8 min, contributing to real-time intraoperative decision-making.

Among the 16 studies assessing efficiency, 13 (81%) reported measurable reductions in time-related outcomes, including operative time, planning time, or fluoroscopy exposure. Absolute time savings ranged from seconds (6–9 s for vascular access) to minutes (3–19 min for operative or procedural duration), and up to ∼50% reductions in specific tasks such as pedicle screw placement time. However, 3 studies reported non-significant or mixed effects, indicating that efficiency gains were context-dependent and less consistent than accuracy outcomes.

A total of 12 [2 RCTs (17%) and 10 observational/feasibility studies (83%)] studies evaluated the impact of immersive technologies—primarily Virtual Reality—on surgical plan modifications (Table 3C). These studies consistently reported changes in preoperative strategies, anatomical assessments, and intraoperative decision-making across various surgical specialties, including neurosurgery, urology, cardiothoracic, and vascular interventions.

VR-based planning frequently led to meaningful alterations in surgical strategies. In skull base and cerebrovascular procedures, VR influenced the choice of head positioning and surgical approach, with statistical significance reported in both skull base meningioma (p < 0.03) and anterior communicating artery aneurysm cases (p < 0.005). Similarly, in pancreatic, hepatic, and renal surgeries, 3D-VR modeling prompted changes in laparoscopic nephrectomy strategies.

VR applications also impacted planning in thoracic and vascular interventions. For example, lung segmentectomy plans were modified in 52% of cases, with accuracy confirmed by pathology in 98% of those revisions. In TAVR procedures, VR insights resulted in alterations in implantation depth, access route, or overall approach in 45.4% of cases. Similarly, target segment selection changed in 40% of video-assisted thoracoscopic surgeries (VATS).

Surgical decision-making also benefitted in robotic-assisted urologic surgeries. Among RALP cases, 32% of surgeons revised their preoperative plans after reviewing VR models. Likewise, 6 out of 15 surgeons modified their approach in robotic-assisted partial nephrectomy following VR review. Surgeons also reported enhanced anatomical understanding and preoperative confidence, with one study noting that 61.1% of decisions were improved and 83.3% of surgeons gained better comprehension through VR visualization.

Across the 12 studies evaluating decision impact, immersive technologies resulted in surgical plan modifications in approximately 30%–52% of cases, with reported rates ranging from 33% to 90% depending on procedure and assessment method. When explicitly quantified, absolute modification frequencies included 6/15 cases (40%), 5/11 cases (45.4%), and 2/6 cases (33%). These findings indicate a meaningful influence on surgical decision-making, although estimates varied widely and were predominantly derived from non-randomized studies.

Across the included studies, several overarching patterns emerged. Virtual Reality was most consistently associated with enhanced spatial understanding, particularly in preoperative planning, where it facilitated accurate anatomical visualization, improved surgical confidence, and often led to modifications in surgical strategies. Augmented Reality, in contrast, demonstrated its greatest strength in intraoperative applications, where real-time overlays contributed to improved surgical accuracy and substantial reductions in operative time. Mixed Reality, although less extensively studied, showed promising benefits in niche applications such as epidural placement, where it reduced both error rates and procedural duration. Despite these common trends, divergences were noted: while some VR-based applications demonstrated significant operative time reductions, others showed nonsignificant or modest effects, highlighting heterogeneity across specialties and procedures. Similarly, AR's intraoperative precision gains were consistently observed, but its added value in preoperative planning was less clear. Collectively, these findings suggest a complementary role of the different immersive technologies—VR strengthening preoperative planning, AR enhancing intraoperative execution, and MR offering targeted benefits in emerging clinical scenarios.