Virtual reality-based 3D prostate models for robot-assisted radical prostatectomy: a prospective exploratory study of membranous urethral length and early urinary continence recovery

Abstract

Objectives

To prospectively explore the feasibility and histopathological concordance of virtual reality (VR)-based 3D prostate models in robot-assisted radical prostatectomy (RARP) and to evaluate the association between VR-derived membranous urethral length (MUL) and early urinary continence recovery.

Methods

VR-based 3D prostate models were generated for 35 patients with localized prostatic adenocarcinoma who underwent RARP. Tumor localization, extraprostatic extension (EPE), and seminal vesicle invasion (SVI) on the 3D models were compared with the final histopathology using Cohen’s kappa coefficient and standard diagnostic accuracy measures. Urinary continence was assessed using the Expanded Prostate Cancer Index Composite (EPIC) survey, with social continence defined as the use of no pads or one safety pad per day and complete continence defined as zero pad use. Continence recovery was evaluated at 1, 3, and 6 months.

Results

The mean patient age in the cohort was 63.2 years, the mean prostate-specific antigen (PSA) concentration was 13.36 ng/mL, and the mean prostate volume was 52.7 cc. Nerve-sparing surgery was performed in 74.3% of the patients, and extended pelvic lymph node dissection was performed in 20%. Pathologic stage ≥T3a occurred in 45.7% of patients, with a Gleason score ≤ 7 in 71.4%, an SVI in 14.3%, and positive margins in 2.9%. Perioperative complications were minimal (5.7%), with no transfusions. Continence recovery occurred in 85.7% of the patients at 3 months and 91.4% at 6 months. MUL correlated negatively with pad use at 3 months (Spearman’s ρ = − 0.52), and a 14-mm MUL cutoff was associated with early continence recovery (AUC = 0.86). VR 3D models showed almost perfect agreement with the final histopathology for extraprostatic extension (EPE) (κ = 0.88; accuracy 94.3%) and perfect agreement for SVI (κ = 1.00).

Conclusions

VR-based 3D prostate models may represent a useful adjunct for surgical planning in RARP and could help inform nerve-sparing strategies. MUL, as measured by these models, may be a valuable factor associated with early urinary continence recovery. These exploratory findings should be interpreted with caution and require confirmation in larger, multicenter prospective studies with longer follow-up periods.

Introduction

Prostate cancer (PCa) is the most frequently diagnosed malignancy in men and the second leading cause of cancer-related death in Western countries [1]. Robot-assisted radical prostatectomy (RARP) has become the most widely performed robotic procedure and is currently regarded as the standard of care for managing localized PCa [2]. Compared with traditional surgical approaches, RARP offers significant advantages, including reduced blood loss, shorter hospital stays, fewer perioperative complications, and lower rates of readmission and reoperation [3]. Despite advances in robotic technology, preoperative planning remains limited by conventional two-dimensional imaging modalities, such as magnetic resonance imaging (MRI), which often lack the anatomical precision necessary for optimal surgical guidance [4]. Recent developments in virtual reality (VR) have enabled the creation of three-dimensional (3D) prostate models using standard radiologic imaging combined with advanced reconstruction software. These VR-based models are gaining traction among clinicians because of their enhanced anatomical visualization [5]. Traditionally, surgeons have relied on the interpretation of multiple imaging planes and the mental reconstruction of 3D anatomy, a process that can be cognitively demanding and imprecise.

Urinary continence following RARP is a critical determinant of postoperative quality of life. Although more than 80% of patients regain continence within one year because of advancements in surgical techniques and anatomical understanding, more than 70% continue to require pads during the early postoperative period [6]. Several preoperative predictors of early continence recovery after nerve-sparing RARP have been identified, and urethral characteristics on MRI, particularly membranous urethral length (MUL), have been identified as key independent predictors of favorable urinary outcomes [7–9].

However, MRI-based measurements of MUL are subject to significant variability, with moderate to poor interreader reliability, particularly between experienced and less experienced readers and across different imaging planes (e.g., coronal vs. sagittal) [10]. In contrast, VR-based 3D reconstructions derived from MRI data may offer more accurate and reproducible visualizations of the prostate, tumor, and surrounding structures, including measurements of MUL along its true anatomical axis, rather than from approximations on orthogonal 2D slices. These models can be integrated into the robotic console during RARP, providing either adjacent or overlaid real-time anatomical guidance [11]. We sought to prospectively explore whether VR-derived 3D prostate models demonstrate clinically acceptable concordance with final histopathology for tumor extension assessment and whether VR-derived membranous urethral length measurements are associated with early urinary continence recovery after RARP. This study was designed as an exploratory feasibility investigation rather than a comparative effectiveness trial. Accordingly, this study evaluated the feasibility and potential utility of VR-based 3D prostate models for preoperative planning in nerve-sparing RARP and investigated the association between MUL and early urinary continence recovery.

Methods

Study design

This prospective single-center study was designed as an exploratory feasibility and concordance investigation evaluating the integration of VR-based 3D prostate models into preoperative and intraoperative surgical workflows during RARP. Between February 2022 and September 2024, patients with localized PCa who underwent nerve-sparing robot-assisted radical prostatectomy (RARP) by a fellowship-trained surgeon (RA) at a tertiary academic medical center were prospectively enrolled in this study. VR-based 3D prostate models were successfully generated in 35/35 attempted cases with histologically confirmed prostatic adenocarcinoma. All procedures were performed using the da Vinci^® Surgical System (Intuitive Surgical, Inc., Sunnyvale, CA, USA).

Preoperative multiparametric magnetic resonance imaging (mpMRI) was performed in all patients to assess extraprostatic extension (EPE) and seminal vesicle invasion (SVI) and to assign PI-RADS scores. A bone scan was also performed to exclude metastatic disease. Patients with imaging evidence of EPE, SVI, pelvic lymphadenopathy, or bone metastases were excluded. Additional exclusion criteria included preoperative urinary incontinence and prior neoadjuvant hormonal therapy.

Urethral catheters were typically removed between postoperative days 7 and 10, unless urine leakage was suspected. Perioperative cystography was not routinely performed unless intraoperative findings or drain output suggested possible leakage. Urinary continence was assessed using a validated item from the Expanded Prostate Cancer Index Composite (EPIC) survey, which evaluates the number of pads or adult diapers used during the day to manage urinary leakage over the preceding four weeks [12, 13]. This question was adapted to assess continence at 1, 3, and 6 months post-surgery.

Prostate 3D image reconstruction and membranous urethral length measurement

Anatomical segmentation was performed using a standardized semiautomated convolutional neural network (CNN)-assisted workflow. Axial T2-weighted 3D turbo spin‒echo (TSE) multiparametric magnetic resonance imaging (mpMRI) volumes were imported into a Python processing pipeline, converted to the NIfTI format, resampled to 0.4 × 0.4 × 3.0 mm voxel spacing, and intensity normalized. A 3D convolutional neural network implemented in PyTorch was subsequently used to generate voxelwise multiclass probability maps for the prostate, seminal vesicles, neurovascular bundles, bladder, urethra, and urethral sphincter. Inference was performed using overlapping 3D patches; probability maps were combined and converted to discrete labels by voxelwise maximum probability assignment, with the final masks resampled back to the native MRI geometry using nearest-neighbor interpolation.

The tumor contours were delineated separately by an experienced radiologist on the mpMRI dataset and incorporated into the 3D reconstruction. All anatomical and tumor masks were reviewed in multiplanar views and manually corrected where necessary before color coding and final 3D meshing. The tumors were color-coded to enhance visual clarity, and the resulting 3D reconstructions were integrated into VR headsets and/or displayed on the da Vinci^® surgical console to support intraoperative navigation. During surgery, the operating surgeon used a console-mounted mouse to manipulate the 3D model manually, facilitating real-time rotation and inspection of regions of interest. Intraoperative feedback on the practicality and utility of the VR models was recorded. Concordance between tumor locations identified on the 3D models and those observed in final histopathological specimens was subsequently assessed.

The membranous urethra was defined as the segment extending from the prostatic apex to the bulb of the corpus spongiosum and was measured on the finalized 3D model (Fig. 1). All MUL measurements were performed by an experienced biomedical engineer who was blinded to the clinical, pathological, and functional outcome data and were independently verified by a second experienced biomedical engineer who was similarly blinded. Discrepancies were resolved by consensus after review. The VR models also enabled three-dimensional visualization of tumor extension beyond the prostatic capsule in relation to surrounding anatomical structures, facilitating an intraoperative assessment of extraprostatic extension (EPE) (Fig. 2).

Fig. 1.

Virtual reality (VR)-based 3D prostate model showing the measurement of membranous urethral length (MUL), which is defined as the segment extending from the prostatic apex to the bulb of the corpus spongiosum. The reconstruction was generated from axial T2-weighted multiparametric MR images using a CNN-assisted semiautomated segmentation workflow. Color coding distinguishes the prostate (light purple), tumor (green), urethra (beige), bladder (yellow), and surrounding anatomical structures

Fig. 2.

Virtual reality (VR)-based 3D prostate model showing the extraprostatic extension (EPE) of the tumor (green) beyond the prostatic capsule

Data collection and outcomes

The demographic data collected for this study included the patient’s age, comorbidities, a history of transurethral resection of the prostate (TURP) or a prior history of cancer, prostate volume, serum prostate-specific antigen (PSA) level, and clinical stage. The recorded operative parameters included the total operative time, estimated blood loss (EBL), need for blood transfusion, length of hospital stay, duration of catheterization, and incidence of perioperative complications. The oncological and functional outcomes assessed included urinary continence, erectile function, pathological stage, and surgical margin status. Postoperative follow-up evaluations were performed at 1, 3, and 6 months.

Definitions and classification criteria

Urinary continence was assessed at 1, 3, and 6 months post-operatively. Social continence was defined as the use of no pads or only one safety pad per 24-hour period and was used as the primary functional endpoint. Complete continence was defined as no pad use over the same period and was reported as a secondary endpoint. A positive surgical margin (PSM) was identified by the presence of cancer cells at the inked surface of the resected sample. Although most patients underwent standard pelvic lymphadenectomy, extended lymph node dissection was selectively performed in patients with high-risk disease features. Biochemical recurrence (BCR) was defined as a postoperative prostate-specific antigen (PSA) concentration > 0.2 ng/mL on two consecutive measurements following an initially undetectable PSA level obtained two weeks after surgery. Postoperative complications were classified according to the Clavien–Dindo system, in which grades I–II were considered minor and grades III–V were considered major complications [14].

Data analysis

The data were analyzed with SPSS version 23.0 for Windows (SPSS Inc., Chicago, IL, USA). Descriptive statistics are presented as frequencies and percentages for categorical variables and as the means ± standard deviations (SDs) or medians with interquartile ranges (IQRs) for continuous variables. Agreement between the VR-3D model findings and final histopathology for extraprostatic extension (EPE) and seminal vesicle invasion (SVI) was assessed using Cohen’s kappa coefficient with 95% confidence intervals. The diagnostic performance of the VR-3D model was further evaluated by calculating the sensitivity, specificity, positive and negative predictive values, and overall accuracy, with histopathology serving as the reference standard. Kappa values were interpreted according to the method of Landis and Koch (≤ 0.20, slight; 0.21–0.40, fair; 0.41–0.60, moderate; 0.61–0.80, substantial; and 0.81–1.00, almost perfect). Continuous variables were compared using the Mann‒Whitney U test or paired t tests, as appropriate. Relationships between continuous variables were evaluated using regression analysis and Spearman’s rank correlation coefficients. A receiver operating characteristic (ROC) curve analysis was performed to identify the optimal membranous urethral length (MUL) cutoff for predicting early urinary continence recovery. A p value < 0.05 was considered to indicate statistical significance.

Results

Thirty-five patients with localized prostatic adenocarcinoma were included in the analysis. The mean age of the patients was 63.2 ± 7.0 years, with a mean prostate volume of 52.7 ± 24.9 cc and a mean PSA concentration of 13.36 ± 7.41 ng/ml. Most patients (85.7%) had clinical stage < T2b disease. Nerve-sparing surgery was performed in 26 patients (74.3%), including 8 patients (22.8%) who underwent bilateral nerve-sparing procedures. Extended pelvic lymph node dissection was performed in 7 patients (20%) (Table 1). Histopathological examination revealed that 16 (45.7%) patients had a pathological stage ≥T3a. A Gleason score of ≤ 7 was observed in 25 patients (71.4%), with a mean cancer volume of 31.8% ± 20.4 in the final pathological specimens. SVI was noted in 5 (14.3%) patients, whereas positive surgical margins were identified in one patient (2.9%) (Table 2).

Table 1.

Baseline and demographic parameters

Parameter (n = 35)		Mean ± SD (range) No (%)
Mean age/years		63.2 ± 7.0
Preoperative morbidities	Hypertension	11 (31.4)
	Diabetes mellitus	16 (45.7)
	Cardiovascular disease	6 (17.1)
Family history of cancer		4 (11.4)
History of TURP		2 (5.7)
Prostate volume (cc)		52.7 ± 24.9 (17–125)
PSA level (ng/mL)		13.36 ± 7.41 (4.76–32.37)
Clinical stage	T1c	26 (74.3)
	T2a	4 (11.4)
	≥T2b	5 (14.3)
PI-RADS score	PI-RADS II	1 (2.9)
	PI-RADS III	3 (8.6)
	PI-RADS IV	18 (51.4)
	PI-RADS V	13 (37.1)
Nerve-sparing (NS) surgery	Unilateral NS surgery	18 (51.4)
	Bilateral NS surgery	8 (22.8)
Lymph node dissection	Limited	28 (80)
	Extended	7 (20)
	Number examined	9.9 ± 5.2 (0–22)

TURP Transurethral resection of the prostate, PSA Prostate-specific antigen, PI-RADS Prostate Imaging–Reporting and Data System, NS Nerve-sparing

Table 2.

Perioperative outcomes and pathological findings

Parameter (n = 35)		Mean ± SD No (%)
Mean console time/min		165 ± 25
Estimated blood loss/ml		99 ± 69
Length of the hospital stay (days)		4.0 ± 0.16
Postoperative PSA level (ng/ml)		0.01 ± 0.06
Pathological findings	Prostate size (cc)	50.8 ± 21.0
	Tumor volume (%)	31.8 ± 20.4 (5–70)
	pT2	19 (54.3)
	pT3	16 (45.7)
	Gleason score of 6	6 (17.1)
	Gleason score of 7	19 (54.3)
	Gleason score of 8	6 (17.1)
	Gleason score of 9	4 (11.4)
	Positive surgical margin	1 (2.9)
	Seminal vesicle invasion	5 (14.3)
	Extraprostatic extension	16 (45.7)

PSA Prostate-specific antigen

Agreement between the findings from the VR 3D model and the final histopathology results was high. Extraprostatic extension was identified in 16/35 patients (45.7%) on histopathology and in 14/35 patients (40.0%) on the VR 3D model. The VR 3D model correctly identified EPE in 14 of 16 affected patients with no false-positive findings (sensitivity 87.5%, specificity 100%, positive predictive value [PPV] 100%, negative predictive value [NPV] 90.5%, accuracy 94.3%; Cohen’s κ = 0.88; 95% CI 0.73–1.00), indicating almost perfect agreement. All five histopathologically confirmed cases of SVI were correctly identified on the VR 3D models, with no false positives (sensitivity 100%, specificity 100%, accuracy 100%; κ = 1.00). An illustrative example demonstrating the concordance between VR-based 3D tumor localization and gross histopathology in a case of extraprostatic extension is shown in Fig. 3.

Fig. 3.

Concordance between gross and histopathological findings of extraprostatic extension. a Gross examination of the prostate specimen showing the area of extraprostatic extension (circled); (b) corresponding histopathology confirming extraprostatic extension at the mid-right posterior zone

Perioperatively, no patients required blood transfusions. Two patients (5.7%) experienced minor complications (Clavien–Dindo grade II). Specifically, one of these patients developed a urinary tract infection, and the other developed a postoperative lymphocele. In accordance with our institutional protocol, routine cystography was not performed prior to catheter removal in any patient; no urine leaks were clinically observed. All patients were instructed to perform postoperative Kegel exercises, specifically 10 repetitions, three times per day.

In sexually active patients, the mean Sexual Health Inventory for Men (SHIM) score improved at 3 months compared with that at the 1-month follow-up (10.3 ± 6.0 vs. 5.9 ± 3.9), although this difference did not reach statistical significance (p = 0.36). This nonsignificant finding likely reflects the small number of sexually active patients with paired data and the considerable interindividual variability in SHIM scores in the early postoperative period. The observed numerical improvement is therefore clinically suggestive but should be interpreted with caution and warrants confirmation in adequately powered cohorts. The overall mean MUL measured on the 3D models was 15.05 ± 7.42 mm (range: 5.6–39.1 mm), and the mean urethral width was 10.7 ± 1.8 mm (range: 6.7–15.1 mm). The MUL did not differ significantly between the bilateral and unilateral nerve-sparing groups (p = 0.07). The borderline p value, despite a numerical difference between groups, may reflect a true small-to-moderate difference that the present sample (n = 35) is underpowered to detect.

Social continence recovery was observed in 30 (85.7%) and 32 (91.4%) patients at 3 and 6 months post-operatively, respectively. Only 3 patients (8.6%) reported persistent incontinence at 6 months after surgery, with an average pad usage of 0.4 ± 1.5 pads per day. The proportion of patients who regained social continence increased significantly from 1 to 3 months post-operatively (14.3% vs. 85.7%, p < 0.001), and the mean number of urinary pads used decreased significantly from 3.4 ± 1.4 at 1 month to 0.8 ± 2.0 at 3 months (p < 0.001).

The membranous urethral length (MUL), as measured on the 3D prostate models, displayed a moderate negative correlation with the number of pads used at 3 months post-operatively (Spearman’s ρ = − 0.52) (Fig. 4). Receiver operating characteristic (ROC) curve analysis identified an optimal MUL cutoff of 14 mm associated with social continence recovery at 3 months, with an area under the curve (AUC) of 0.86, a sensitivity of 89%, and a specificity of 68% (Fig. 5).

Fig. 4.

Spearman’s rank correlation analysis between the membranous urethral length (MUL) measured on the VR-based 3D prostate model and the number of pads used per day at 3 months post-operatively (ρ = − 0.52). Each point represents an individual patient (n = 35). A longer MUL was associated with fewer pads needed

Fig. 5.

Receiver operating characteristic (ROC) curve for the predictive value of the preoperative VR-derived membranous urethral length (MUL) for the recovery of social continence at 3 months post-operatively. The optimal MUL cutoff was 14 mm; the area under the curve (AUC) was 0.86; the sensitivity was 89%; and the specificity was 68%

Discussion

Postoperative urinary continence is influenced by the integrity of both the internal urethral sphincter, which is composed of smooth muscle fibers extending from the bladder neck to the perineal membrane, and the external urethral sphincter, which consists of striated muscle fibers concentrated around the membranous urethra [15]. Preservation of these anatomical structures during RARP is essential for facilitating early recovery from continence. Post-RARP urinary incontinence is influenced by several patient-related factors, including age, preexisting urinary symptoms, prior prostate surgeries, body mass index (BMI), and surgical factors, such as the use of a nerve-sparing approach [16]. Preoperative prostate MRI parameters have been proposed to predict urinary incontinence risk following radical prostatectomy. However, conventional two-dimensional MRI often lacks the resolution and accuracy necessary to capture fine anatomical details, partly because of variability in measurement techniques and low interobserver agreement [17].

The advent of 3D visualization technologies has enabled a more precise representation of anatomical structures, ushering in an era of personalized surgical planning. VR-based 3D prostate model reconstruction is increasingly recognized as a valuable tool for enhancing surgeons’ understanding of the surgical field. Significant advancements in 3D modeling have expanded its application in the context of minimally invasive robotic surgery for both kidney and prostate tumors [18, 19]. In the context of RARP, VR-based 3D models have been increasingly investigated as adjunctive tools for surgical planning, nerve-sparing decision-making, anatomical visualization, and intraoperative navigation during RARP. The development of these VR-3D models requires high-quality bidimensional digital imaging and communications in medicine (DICOM) images, which are manually segmented to reconstruct relevant anatomical structures. These virtual models can be formatted for compatibility with various devices, including VR headsets and 3D printers [20].

A recent multicenter, randomized, single-blind study demonstrated that 3D VR model-assisted surgical planning improved oncological outcomes without compromising functional results. Notably, the surgical plan for 32% of patients was modified on the basis of the findings obtained from the VR model, and a significant increase in bilateral nerve-sparing procedures was identified [21]. Meta-analyses have confirmed that a longer preoperative MUL, as measured using MRI, is an independent predictor of early urinary continence recovery within one month and remains a prognostic factor at 12 months [22]. Our findings are consistent with previous reports describing the association between preoperative MUL and urinary continence. The observed comparable correlation between our preoperative 3D model and histopathology with respect to SVI and EPE, without increased positive surgical margins, suggests potential utility for intraoperative navigation from an oncological standpoint that warrants further validation.

In another study, patients who underwent bilateral nerve-sparing techniques had significantly longer postoperative MUL on MRI than patients who did not undergo such procedures [23]. A longer baseline MUL is associated with greater preservation of smooth muscle fibers, which improves the urethral pressure profile and facilitates continence recovery [22]. Several studies have proposed specific MUL cutoff values to predict urinary continence outcomes. For instance, Ikarashi et al. suggested a threshold of 12 mm for predicting continence at 3 months post-RARP, whereas our exploratory analysis identified a cutoff of 14 mm, with a sensitivity of 89% and specificity of 68% (compared with 80% and 70%, respectively, in that study) [23].

In the present cohort, 19 patients (54.3%) had an MUL ≥ 12 mm, and 13 patients (37.1%) had an MUL ≥ 14 mm. Notably, all five patients who regained social continence within the first month had an MUL ≥ 12 mm. The VR-derived cutoff of 14 mm demonstrated favorable predictive performance within this exploratory cohort; however, direct comparisons with conventional MRI-derived MUL measurements were beyond the scope of the present study. This 14-mm threshold should therefore be regarded as an exploratory, hypothesis-generating finding rather than a clinically actionable decision-making tool, and requires validation in larger external cohorts before any clinical implementation. Additionally, compared with only one patient in the persistently incontinent group, 77% of patients who achieved early postoperative continence had an MUL ≥ 14 mm. Although comparative studies between VR and 3D printing remain limited, VR-based models may have promising applications in surgical education, training, and patient counseling because they may facilitate an intuitive visualization of the anatomy of the diseased tissue and treatment plans [24].

Although patients with overt imaging evidence of EPE or SVI were excluded preoperatively, pathological upstaging after apparent organ-confined disease on mpMRI remains a well-recognized phenomenon in prostate cancer surgery. Contemporary studies have demonstrated that mpMRI has limited sensitivity for detecting microscopic EPE and SVI despite good specificity, resulting in clinically occult locally advanced disease in a subset of patients undergoing radical prostatectomy. Because the VR-based reconstructions in the present study were derived directly from mpMRI datasets, their performance is inherently dependent on the quality and staging limitations of the source imaging. Accordingly, the pathological upstaging observed in this cohort should be interpreted within the known diagnostic limitations of contemporary mpMRI rather than as evidence of being unique to the VR platform.

However, the effect of MUL on postoperative continence should be interpreted with caution, as variations in continence definitions, measurement methods, study designs, and sample sizes complicate direct comparisons. The additional use of the MUL involves preoperative counseling. Creating a validated 3D VR model can serve as a useful tool to explain a patient’s risk of postoperative urinary incontinence. Larger studies are needed to create validated numbers for the MUL; however, this process is already helpful as a visual counseling tool for the patient and for setting the expected timeline [25].

This study has several limitations that should be acknowledged. First, the relatively small sample size (n = 35) of patients from a single center limits the statistical power and generalizability of the findings. Although a prospective design increases internal validity, larger multicenter studies are needed to confirm these results. Second, with only five patients who were persistently incontinent at 3 months, the cohort falls well below the conventional events per variable threshold required for stable multivariable logistic regression. We therefore restricted the analysis to univariate associations to avoid overfitting and inflated effect estimates. As a result, the observed association between MUL and early continence recovery should be interpreted as a univariate signal rather than an independent predictor; adequately powered multivariable analyses adjusting for age, BMI, prostate volume, and nerve-sparing status are needed in larger cohorts. Third, follow-up for continence outcomes was limited to 6 months. Although early continence is itself a clinically meaningful endpoint that closely correlates with patients’ quality of life during the first postoperative year, the conventional reference time point for a definitive continence status is 12 months. Longer follow-up is needed to confirm whether the observed MUL threshold predicts ultimate continence recovery. Fourth, the present study did not include direct head-to-head comparisons between VR 3D and conventional 2D MRI-based MUL measurements or tumor extension assessments. Importantly, the present study was not designed to establish the superiority of VR-based modeling over conventional MRI interpretation or standard surgical planning. Rather, the objective was to evaluate the feasibility, histopathological concordance, and potential utility of VR-derived anatomical assessment as a hypothesis-generating platform for future comparative investigations. While indirect evidence, including the documented limitations of 2D MRI interreader reliability and the demonstrated impact of 3D VR models on intraoperative decision-making in randomized trials, supports incremental value, a prospective comparative analysis is needed to definitively establish the superiority of 3D VR over conventional MRI. Fifth, although all MUL measurements were performed by two independent, blinded, experienced biomedical engineers with a consensus resolution of discrepancies, formal interobserver and intraobserver reliability analyses were beyond the scope of this study. Establishing the reproducibility of 3D-derived VR measurements across multiple readers and centers represents an important focus of future validation work. Sixth, the implementation of VR-based models requires advanced imaging, software infrastructure, and technical expertise, which may limit their widespread adoption in resource-constrained settings. Finally, while this study describes the concordance between VR-derived tumor localization and final histopathology, the clinical effects of VR-guided surgical planning on positive surgical margin rates and functional recovery must be further validated in adequately powered randomized controlled trials. Nonetheless, the findings of this research support the potential of MUL, as measured using 3D VR models, to serve as a valuable factor associated with early continence recovery following radical prostatectomy, particularly given the limited availability of widely accessible MRI-based 3D modeling tools.

Conclusions

In this exploratory study, virtual reality-based 3D prostate models may represent a useful adjunct for assessing tumor extension and could help inform surgical decision-making during robot-assisted radical prostatectomy. Membranous urethral length, as measured using these models, may be a valuable factor associated with early urinary continence recovery. However, these exploratory findings derived from a single-center cohort of 35 patients with a 6-month follow-up should be interpreted with caution and require further evaluation in larger, multicenter prospective studies with longer follow-up and adequately powered multivariable analyses.

Abbreviations

RARP

Robot-assisted radical prostatectomy

MUL

Membranous urethral length

VR

Virtual reality

3D

Three-dimensional

2D

Two-dimensional

MRI

Magnetic resonance imaging

mpMRI

Multiparametric magnetic resonance imaging

EPE

Extraprostatic extension

SVI

Seminal vesicle invasion

PSA

Prostate-specific antigen

AUC

Area under the curve

ROC

Receiver operating characteristic

EBL

Estimated blood loss

TURP

Transurethral resection of the prostate

CNN

Convolutional neural network

DICOM

Digital Imaging and Communications in Medicine

BMI

Body mass index

Authors’ contributions

R.A.A. conceived and designed the study, performed the surgical procedures, and supervised the project. A.M.A. contributed to patient recruitment and data collection. R.A.A and A.A. drafted the manuscript. S.S. contributed to study coordination and manuscript review. M.E. performed data analysis and contributed to interpretation of the results. All authors reviewed and approved the final manuscript and agree to be accountable for all aspects of the work.

Funding

This research received no external funding.

Data availability

The datasets generated and analyzed during the current study are not publicly available owing to institutional data-protection policies and patient privacy considerations but are available from the corresponding author upon reasonable request.

Declarations

Ethics approval and consent to participate

This study was approved by the Unit of Biomedical Ethics Research Committee, Faculty of Medicine, King Abdulaziz University, Jeddah, Saudi Arabia (Reference No. 154 − 20). Written informed consent was obtained from all participants. All procedures performed in this study were conducted in accordance with the relevant guidelines and regulations, including the principles of the Declaration of Helsinki and its later amendments.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.