Are 3D Image Guidance Systems Ready for Use? A Comparative Analysis of 3D Image Guidance Implementations in Minimally Invasive Partial Nephrectomy

Abstract

Introduction:

Three-dimensional image-guided surgical (3D-IGS) systems for minimally invasive partial nephrectomy (MIPN) can potentially improve the efficiency and accuracy of intraoperative anatomical localization and tumor resection. This review seeks to analyze the current state of research regarding 3D-IGS, including the evaluation of clinical outcomes, system functionality, and qualitative insights regarding 3D-IGS's impact on surgical procedures.

Methods:

We have systematically reviewed the clinical literature pertaining to 3D-IGS deployed for MIPN. For inclusion, studies must produce a patient-specific 3D anatomical model from two-dimensional imaging. Data extracted from the studies include clinical results, registration (alignment of the 3D model to the surgical scene) method used, limitations, and data types reported. A subset of studies was qualitatively analyzed through an inductive coding approach to identify major themes and subthemes across the studies.

Results:

Twenty-five studies were included in the review. Eight (32%) studies reported clinical results that point to 3D-IGS improving multiple surgical outcomes. Manual registration was the most utilized (48%). Soft tissue deformation was the most cited limitation among the included studies. Many studies reported qualitative statements regarding surgeon accuracy improvement, but quantitative surgeon accuracy data were not reported. During the qualitative analysis, six major themes emerged across the nine applicable studies. They are as follows: 3D-IGS is necessary, 3D-IGS improved surgical outcomes, researcher/surgeon confidence in 3D-IGS system, enhanced surgeon ability/accuracy, anatomical explanation for qualitative assessment, and claims without data or reference to support.

Conclusions:

Currently, clinical outcomes are the main source of quantitative data available to point to 3D-IGS's efficacy. However, the literature qualitatively suggests the benefit of accurate 3D-IGS for robotic partial nephrectomy.

Introduction

Minimally invasive nephron sparing surgery (partial nephrectomy) has been adopted as the standard of care for small renal masses when surgically feasible.¹ There are many advantages to a laparoscopic/robotic minimally invasive approach to partial nephrectomy, including reduced postoperative pain, shorter recovery times and hospital stays, and improved cosmesis.² Robotic minimally invasive surgery has additional advantages of stereoscopic vision and greater instrument dexterity when compared to laparoscopic procedure.³ Despite these advantages, minimally invasive surgery is not without challenges. Minimally invasive operation lacks the traditional handling and feedback of an open procedure. This makes tissue navigation and vessel identification more challenging. Furthermore, performing a partial kidney resection requires the surgeon to keep an accurate mental map as the procedure progresses.

While operating, the surgeon must recall the patient's two-dimensional (2D) preoperative scans, mentally generate a rough 3D estimate of the anatomy, and align that estimate with the operative field. Currently, surgeons remove renal masses using only direct visualization as guidance for dissection and resection. This may prolong procedure time and risks injury to the surrounding tissue. Assistance with intraoperative visualization of the patient's anatomy remains a need. Tools are now being developed for image-guided surgery (IGS), which can potentially assist the surgeon during minimally invasive surgery for kidney cancer. The purpose of this contemporary literature review is to highlight these developing technologies, evaluate the reported outcomes, and bring attention to areas that need further study.

Background

Current clinical technology

Currently, there are relatively few routinely used tools to assist surgeons during minimally invasive partial nephrectomy (MIPN). Near-infrared fluorescence (e.g., indocyanine green) can be injected intravenously and activated with a light-emitting diode. This is typically used to identify areas of perfusion as it flows through the vasculature.⁴ The da Vinci surgical system can switch between white light and fluorescent views to enable the surgeon to better identify the boundary between tumor and kidney. However, fluorescent dye injection is not yet part of the standard for renal cancer treatment.⁵

In recent years, surgeons have integrated laparoscopic ultrasound probes into their procedures as another method to gain information. In this study, the surgeon can hold the probe with a laparoscopic or robotic tool and sweep the fat/kidney surface to reveal hidden vessels and the tumor-kidney boundary. Unlike fluorescence, ultrasound does not require injection and still provides useful information.^6,7 Although both techniques help surgeons localize the tumor-parenchyma boundary and relevant vessels, they still require the surgeon to keep a cognitive map of the patient's preoperative imaging-based anatomy. As an alternative, clinicians and researchers have been working to develop tools to integrate preoperative/intraoperative relevant anatomical data from scans directly into the procedure performance.

Active image guidance research

To this end, two sectors of image guidance technology are being developed. The first involves the fusion of data obtained in real time through intraoperative imaging modalities (CT, MRI, ultrasound, etc.) into the surgeon's field of view. Currently, this image guidance technique is used for ablative procedures.^8,9 Although these techniques give detailed, live information to the proceduralist during ablation, they require the use of expensive equipment that is not readily accessible and also lacks the accurate 3D information necessary for IGS.¹⁰

The second image guidance method involves constructing a three-dimensional model of the patient's anatomy from preoperative imaging scans. This model can be accessed by the surgeon during model generation to enable a preoperative view of the patient's anatomical structures. Furthermore, the model is displayed to the surgeon during the procedure, to provide additional spatial and anatomical information in real time. The 3D model is presented to the surgeon in a separate display (e.g., Intuitive's TilePro),¹¹ or it is overlaid onto the anatomy through an augmented reality approach.¹²

Most three-dimensional image-guided surgery (3D-IGS) implementations are designed for use during robot-assisted partial nephrectomy (RAPN). There are, however, instances of 3D-IGS being used during laparoscopic partial nephrectomy.¹³ Regardless of approach, 3D-IGS systems comprise four main components: (1) Image acquisition/model generation, (2) model-to-surgical scene registration, (3) three-dimension tool tracking, and (4) display of model and tool position.¹⁴

3D-IGS Components

Image acquisition and model generation

Preoperative staging imaging can be used to generate 3D models of relevant anatomy. Typically, each anatomical structure is manually segmented by switching between coronal, sagittal, and axial views. The manual model generation process can take more than 48 hours for completion.¹⁵ Furthermore, research has suggested that the manual segmentation process can introduce variability and error in the subsequent 3D model.¹⁶ As an alternative, automatic segmentation algorithms have been proposed to increase the efficiency and accuracy of 3D model generation.^17,18 In addition to variation in segmentation, studies have demonstrated as much as a 46 mm shift of the kidney between supine and flank position.¹⁹ The discrepancy between preoperative scanning and intraoperative positioning could affect the accuracy of the 3D model when applied to the surgical scene.

Model-to-operative field registration

Model-to-operative field registration is a critical step in the image guidance process, with perhaps the most variation among 3D-IGS systems. This registration is required to determine the correct orientation and location of the 3D model with respect to the operative field, that is, the transformation between the model and physical space. This transformation can be estimated manually, or it can be computed through various processes.

When done manually, a technician continually aligns the 3D model with the surgical scene through the use of a space mouse or other similar 3D input device.²⁰ Alternatively, the surgeon can align the model according to their needs through a preoperative surgical planning system,¹³ or within an input device connected to the da Vinci surgeon console.^21,22 Manual registration implementations are thought to be accurate enough to improve surgical perception.²³ Further, researchers cite this method as an avenue to overcome intraoperative tissue deformation, as constant adjustments from the technician seek to capture the effects of deformation.^24,25 Conversely, manual alignment's requirement for constant input from a technician or surgeon subjects it to human error and latency.

Computed registration can be performed in multiple ways. Point-based registration involves matching distinct points in both model and physical space. The points can consist of anatomical features or extrinsic fiducial markers.¹⁴ Feature-based registration is not ideal for the kidney, because the anatomy lacks unique features and geometries. To overcome this challenge, researchers have demonstrated point-based registration through the implantation of extrinsic fiducial markers. After marker implantation, the kidney is imaged, and the markers are localized within the scans. Registration is then achieved by computing the transformation between markers in the image space and markers in the physical space.^23,26,27 Although accurate, this registration method requires intraoperative imaging and the use of implanted markers, which add to the overall patient morbidity. Other methods for point-based registration have also been introduced, such as collecting the necessary points with a laser or tracked holographic probes.^27,28

These techniques require additional equipment that is currently not compatible with minimally invasive approaches. Surface-based registration is an alternative to point based. With this registration method, the surface of the kidney in physical space is matched to the surface of the kidney in image space. The collection of surface points in a minimally invasive setting is challenging, and one group has proposed to use the robotic tools already inside the patient to gather the necessary surface tracing data.¹¹ This technique is useful in the case of RAPN since there are significant challenges in intraoperative tracking of traditional laparoscopic instruments. Surface registration and segmentation have also been performed by using computer vision.²⁹ This method only requires endoscope video feed to perform the registration, but it is computationally intensive and not fully robust to deformation, clinical variations, and intraoperative visual obstructions to the surgical field such as smoke or blood.

Another distinction can be made among registration methods. If the 3D modeled organ only translates and rotates, but does not deform, the registration is considered rigid. Rigid registrations are common among 3D-IGS systems for many areas of the body. For operation involving soft tissue such as the kidney, deformation can cause registration error when using a rigid model. Clamping the vessels and excising the tumor is known to cause deformation and structural changes, respectively, to the shape of the kidney.³⁰ Despite this, many 3D-IGS systems for partial nephrectomy employ a rigid registration method.^{15,19

–26,31

–37}

This method generally provides reasonable accuracy, but it does not fully capture the motion of the anatomy throughout the procedure. As an alternative, deformable registration models can be implemented. This allows the surface geometry of the 3D model to change with the anatomy throughout the procedure.^{29,30,38

–41} Development and evaluation of deformable registration will be a necessary area of research for 3D-IGS during partial nephrectomy.

Tool tracking and display of model and tool position

To relate the current operative field to the 3D image guidance scene, the surgical tools must be tracked. This enables accurate depiction of the tools within the image guidance display. Furthermore, it is necessary to collect data points for systems that deploy surface-based registration. Possible tracking methods include triangulation through optical, magnetic, and sonic tracking, or geometric localization with robotic encoders. Often, 3D-IGS researchers opt out of magnetic tracking owing to the many ferromagnetic and electromagnetic objects within the operating room. Optical tracking is the most common method used in 3D-IGS, as it is applicable for both laparoscopic and robotic procedures. Optical tracking requires an uninterrupted line of sight, which can often be obstructed with the dynamic surgical scene and tools.¹⁴ The type of tracking necessarily depends on the other components of the 3D-IGS being implemented.

After optimal tracking is achieved, the 3D model and tools can be displayed to the surgeon. Often, the information is overlaid onto the endoscopic feed in an augmented reality manner (Fig. 1a) (see e.g., Chen et al.,²⁰ Michiels et al.,²³ Pratt et al.,³² Chauvet et al.³⁹), or it is presented in a separate display alongside the endoscope video (Fig. 1b) (see e.g., Amparore et al.,¹⁵ Hughes-Hallett et al.,²¹ Porpiglia et al.,²⁵ Amparore et al.,³¹ Siebold et al.³⁷). With the implementation of these new inputs, there has been some concern raised regarding the split attention required to utilize 3D-IGS and its effects on surgical outcomes.⁴² Surgeons must be aware of the possibility for inattention blindness, but there is currently no hint of significant impairment.

FIG. 1.

(a) Augmented reality IGS display. (b) Side-by-side IGS display. IGS, image-guided surgery.

3D-IGS Evaluations

Several 3D-IGS systems have been tested to determine their effectiveness. Often, when employed in a clinical setting, researchers will compare their system's performance to the standard of care. Typical clinical data reported include warm ischemia time, instances of collecting system violation, operating time, tumor margins, and so on. These clinical results are important when evaluating 3D-IGS's overall effect on the procedure, but researchers also report results regarding the functioning of the system itself. Notable nonclinical results include registration accuracy (see e.g., Teber et al.,²³ Pratt et al.,³² Zhang et al.⁴¹), image segmentation time (see e.g., Nosrati et al.²⁹), and organ shift (see e.g., Schneider et al.¹⁹). To further illustrate the effectiveness of their systems, researchers report qualitative statements regarding 3D-IGS's influence on the procedure.

Methods

We systematically reviewed the clinical literature regarding relevant 3D image guidance systems for MIPN. PubMed and Web of Science were searched with the terms “partial nephrectomy” and “image guidance” to find relevant studies published up to October 3, 2022. Titles and abstracts were screened to determine relevance. Once initial screening was complete, the articles were read to determine if they satisfied the search criteria. Studies were excluded if they contained non-3D guidance models such as fluorescence guidance, evaluated procedures other than partial nephrectomy (e.g., cryoablation), or only evaluated performance on phantom models.

Literature search methodology

This systematic review followed the narrative synthesis methodology established by Popay et al.⁴³ We categorized the resulting literature based on registration method, the result types reported, and the stated limitations of each system. Beyond standard systematic review techniques that produce a collection of references, we also conducted a qualitative analysis of the resulting literature using the Dedoose software, which enables authors to collaboratively identify themes in the included texts.⁴⁴ In the qualitative portion of the analysis, we investigated the perceived 3D-IGS effectiveness and the researcher/surgeon confidence in the system, including instances in which 3D-IGS prevented possible adverse events.

We followed an inductive coding approach to assign keywords to passages in the studies, which share underlying meaning.⁴⁵ Independent codes were developed by two authors (P.C. and S.K.G.) to analyze one study from the set of published articles that resulted from the systematic review described above. Then, the individually produced codes were reviewed and discussed to produce the final set of codes, a code book, agreed upon by the authors. Agreement on the codes in this initial study's analysis was over 90%. The final set of codes was applied to the remaining studies by one author.

Study Quality Assessment

Two authors (P.C. and S.S.) independently conducted a quality assessment of studies included in this systematic review using the Mixed Methods Appraisal Tool (MMAT).⁴⁶ The MMAT is an established assessment tool that enables the methodological evaluation across five different study categories (randomized controlled trials, nonrandomized controlled trials, quantitative descriptive studies, qualitative studies, and mixed methods studies), making it suitable for this review. In adherence with the MMAT guidelines, we present a textual description of the quality assessment in our Results section, rather than calculate an overall study quality score. Furthermore, no study was excluded from this review based on the outcomes of this assessment. After the independent assessment, the two authors discussed their results and resolved any discrepancy such that the final agreement reached 100%.

Results

The initial literature search resulted in 209 studies. After removing studies based on the exclusion criteria and removing duplicates, 25 studies remained (Fig. 2). Key characteristics of these 25 remaining studies are presented in Table 1.

FIG. 2.

Study selection flow chart.

Table 1.

Key Study Characteristics

Citation key	Authors	Year	Study type	Number of subjects	Registration method
⁴⁷	Baumhauer et al.	2008	Animal	3	Point based, Rigid
²³	Teber et al.	2009	Animal	10	Point based, Rigid
²³	Teber et al.	2009	Human	10	Manual, Rigid
³⁰	Altamar et al.	2011	Animal	6	NA
³⁰	Altamar et al.	2011	Human	Unclear	Surface based, Rigid
³²	Pratt et al.	2012	Human Images	1	Manual, Rigid Semiautomated
³²	Pratt et al.	2012	Human	3	Manual, Rigid Semiautomated
¹⁹	Schneider et al.	2013	Human Images	10	Surface based, Rigid
³⁶	Friets et al.	2013	Animal	1	Surface based, Rigid
²¹	Hughes-Hallett et al.	2014	Human	5	Manual, Rigid
²⁰	Chen et al.	2014	Human	15	Manual, Rigid
²⁴	Wang et al.	2015	Human	44 (21 with 3D-IGS and 23 without, 14 selected for control)	Manual, Rigid
¹⁶	Hughes-Hallett et al.	2016	Human Images	10	NA
²⁶	Simpfendörfer et al.	2016	Animal	1	Point based, Rigid
²⁶	Simpfendörfer et al.	2016	Human	10	Point based, Rigid
²⁹	Nosrati et al.	2016	Animal	10	Surface based, Deformable
²⁹	Nosrati et al.	2016	Human	5	Surface based, Deformable
¹³	Wang et al.	2017	Human	94 (49 with 3D-IGS and 45 without)	NA
⁴⁰	Kong et al.	2017	Animal	10 ex vivo and 1 in vivo	Point based, Deformable
³⁸	Edgcumbe et al.	2018	Phantom	32 (16 with PARIS and 16 with only LUS)	Surface based, Deformable
³⁹	Chauvet et al.	2018	Phantom	NA	Surface based, Deformable
³⁹	Chauvet et al.	2018	Animal	62 (29 with 3D-IGS and 33 without)	Surface based, Deformable
⁴⁸	Wake et al.	2019	Human Images	20	NA
⁴¹	Zhang et al.	2019	Phantom	NA	Surface based, Deformable
⁴¹	Zhang et al.	2019	Human	1	Surface based, Deformable
³⁴	Kokko et al.	2020	Human Images	1 case recording, 5 study subjects for segmentation	Manual, Rigid Semiautomated
²⁵	Porpiglia et al.	2020	Human	91 (48 with 3D-IGS and 43 with 2D US)	Manual, Rigid
³⁷	Siebold et al.	2020	Human Images	1 set of scans	Surface based, Rigid
²²	Michiels et al.	2021	Human	314 (157 with 3D-IGS and 157 without)	Manual, Rigid
¹⁵	Amparore et al.	2022	Human	222 (79 with 3D-IGS and 143 without)	Manual, Rigid
³³	Kokko et al.	2022	Human Images	1	Surface based, Rigid
³¹	Amparore et al.	2022	Human	3	Manual, Rigid

“Human” study type denotes real-time IGS implementation during a procedure, whereas “Human Images” denotes implementation outside of a clinical procedure. “Semiautomated” registration denotes a combination of both manual and computed alignment.

2D = two dimensional; AR = augmented reality; LUS = laparoscopic ultrasound; PARIS = projector-based augmented reality intracorporeal system; US = ultrasound.

A majority (15 out of 25) of the included studies were assessed with the MMAT criteria in the “Quantitative descriptive” category.^{19,20,23,26,30

–34,36,37,40,41,47,48} The studies in this category present results of a novel methodology to address their respective research questions, as applied to a single study population without comparison to a “control” group. These studies present new and developing methodologies, focusing on feasibility and initial analysis. MMAT criteria 4.4, addressing the risk of nonresponse bias, was not considered for studies in this category, which did not include human subjects.

Accounting for this criterion exclusion, the studies in this category satisfied the criteria, but had generally small sample sizes, indicative of the exploratory nature of these studies. A smaller portion of included studies (7 out of 25) were assessed in the “Quantitative nonrandomized” criteria category.^{13,15,16,22,24,25,38} The studies in this category satisfied the MMAT criteria, but frequently stated the nonrandomized nature of studies as a limitation. Note further discussion of all included study limitations is presented in the “Nonclinical Results” section below.

One study was assessed with the “Quantitative randomized controlled trial” MMAT category.³⁹ Although this study satisfied the MMAT criteria, it should be noted that the study subjects were porcine, and the overall outcomes of the study would be better generalizable if the methodologies were assessed in humans. The remaining two studies were assessed using multiple MMAT criteria, as they each employed mixed methods within the article. One was assessed with the “Qualitative,” “Quantitative descriptive,” and the “Mixed methods” categories.²⁹ The remaining study was assessed with the “Quantitative randomized controlled trials,” “Quantitative descriptive,” and “Mixed methods” categories.²¹ All respective criteria were satisfied, except for MMAT criteria 5.1, which regard the inclusion of explicit rationale for using a mixed methods study design.

Clinical results

Of the 25 studies in our systematic literature review, 8 presented clinical study results, including median operation time, median warm ischemia time, final pathology, and excision margins.^{13,15,20,22,24
–26,31} Unless stated otherwise in what follows, no significant difference in specific outcomes was found between 3D-IGS and non-3D-IGS patients. In work done by Wang et al., 3D-IGS was implemented in all 49 patients.¹³ In their study, Wang et al. found the 3D-IGS patients required less operative time (159.0 vs 193.2 minutes; p < 0.001) and had less estimated blood loss (148.1 vs 176.1 mL; p < 0.001).²⁴ Another study by the same group reported negative margins and no recurrence across all 3D-IGS patients.²⁰ One study found the implementation of 3D-IGS added 15.5 minutes to the overall operation time.²⁶

In Amparore et al.'s study, 3D-IGS patients demonstrated better postoperative outcomes in terms of estimated glomerular filtration rate (eGFR) (17.7% vs 22.2%; p = 0.03), postoperative complications (16.5% vs 23.1%; p = 0.03), and major complications (Clavien Dindo >III, 2.5% vs 5.6%; p = 0.03).¹⁵ They also found that 3D-IGS assistance predicted higher rates of partial nephrectomy (odds ratio = 1.42; p = 0.03). In a separate study by the same group, 3D-IGS patients were found to have a lower rate of global ischemia (45.8% vs 69.7%; p = 0.03), higher rate of enucleation (62.5% vs 37.5%; p$ = $0.02), and fewer collecting system violations (10.4% vs 45.5%; p = 0.003).²⁵ This study also found the use of 3D-IGS to correlate to a “low risk” of procedure-related complications, plus the 3D-IGS patients lost less estimated renal plasma flow at their 3-month follow up (−12.38 vs −18.14; p = 0.01).

This group also demonstrated the use of their 3D-IGS system in three challenging clinical cases, which involved high-complexity renal masses. In this study, researchers report successful surgical outcomes for all three patients.³¹ In another study, researchers found the major complication rate to be lower for 3D-IGS patients (3.8% vs 9.5%; p = 0.04).²² They also found the 3D-IGS patients' eGFR variation to be lower at follow-up (−5.6% vs −10.5%; p = 0.002), and the 3D-IGS patients' trifecta achievement rate (negative cancer margin, minimal renal functional decrease, and no urological complications), as described by Hung AJ et al.,⁴⁹ to be higher (55.7% vs 45.1%; p = 0.005).

Nonclinical results

Registration

Model-to-operative field registration is challenging and critical to any 3D-IGS system. A graphical comparison of registration methods used in articles identified by our systematic review can be found in Fig. 3. Manual registration methods are most used across all included studies (48%; Fig. 3) and most used among the subset of those studies using real-time clinical data (57%; Fig. 3). Furthermore, across all studies, surface-based registration was used more often than point-based registration (nine uses vs five uses). One study employed a computed registration method for their phantom experiments, and then used a manual registration for their clinical experiments.²³ Another study by means of real-time clinical data used their 3D-IGS system exclusively for preoperative planning, so intraoperative registration was not necessary.¹³

FIG. 3.

Registration methods employed across all studies (left) and the real-time clinical studies (right) in our systematic review.

3D-IGS limitations

Limitations cited by each study are listed in Table 2. Each limitation is as stated by the respective authors themselves, and then grouped into overarching categories. Registration obstacles accounted for most limitations found across all studies, with study/system design being the second most common limitation. Among the studies that cited registration as an obstacle, soft tissue deformation was mentioned most often as the primary registration challenge. Three studies acknowledged the need for larger scale studies with randomized samples. In addition, four studies attributed the model generation (segmentation) aspect of their system to be limiting.

Table 2.

Limitations Described by the Authors in Each Study

Obstacle		Study
Registration	Soft tissue deformation	^{19,21,23,24,30,32,34}
	Manual registration	^{15,20,21,24,31}
	Computed registration	^{29,33,37,40,47}
Study/system design	Split view between endoscope and TilePro	²¹
	Limited/nonrandom sample size	^13,15,18,25
	Financial cost	³¹
	Inclusion of high time RAPN surgeons beyond their learning curves	²²
	Radiation dose (CBCT)	²⁶
2D to >3D	AR 2D to >3D reconstruction	²³
	Building time for 3D models (10+hours)	^15,31
	Inaccurate segmentation	¹⁶
Results	Unclear significance of errors reported	⁴⁸

CBCT = cone beam computed tomography; RAPN = robot-assisted partial nephrectomy.

Registration method vs data type reported

Two important indicators for 3D-IGS system efficacy are the relevant clinical outcomes and the improvement to the surgeon's accuracy. The frequency of these reported result types was quantified across all the studies in our systematic review, which used in vivo registration (Fig. 4). Six of the eight manual registration clinical studies reported both clinical outcomes and qualitative assessment of surgeon accuracy. The other two reported only qualitative surgeon accuracy statements. Of the four in vivo studies utilizing a computed registration method, two reported qualitative surgeon accuracy improvement. Furthermore, one of those two studies also reported patient characteristic and surgical outcome data. No study reported a quantitative assessment of surgeon accuracy.

FIG. 4.

Registration method vs result type reported.

Qualitative results

A qualitative analysis of the clinical studies, comparing 3D-IGS to the standard of care in MIPN, in our systematic review (9 of the 25 articles) led to the identification of six themes. The identified themes describe the necessity of 3D-IGS, 3D-IGS's ability to improve surgical outcomes, confidence in the systems, instances of enhanced surgeon ability/accuracy, anatomical explanations for qualitative system assessment data, and subjective statements, which convey expert opinion rather than quantitative evaluations for support. Each theme is presented in Table 3.

Table 3.

Themes Identified in Clinical Studies Comparing Three-Dimensional Image-Guided Surgery to the Standard of Care

Code	Definition	Applicable sample quotes	Reference
Image guidance is necessary	Image guidance should be implemented to improve outcomes	Over the last decades, the indication for minimal invasive procedures has expanded to endophytic and also more complex tumors (hilar, T2–T3). In these cases, surgeons are challenged by the precise analysis of available imaging to understand tumor location and spreading before and during operation. Particularly in selective segmental artery clamping, an exact visualization of arterial branches during operation is of paramount interest to anticipate dissection strategies.	²⁶
Image guidance is necessary	Image guidance should be implemented to improve outcomes	The management of intrarenal tumors presents a technical challenge during laparoscopic partial nephrectomy because these tumors cannot be visualized on the kidney surface and external visual cues to guide tumor localization or margins are lacking.	²⁰
Image guidance improved surgical outcomes	The use of image guidance improved surgical outcomes	In this largest series to date reporting 3D-IGRAPN, compared to a high-volume surgeon RAPN cohort, our results showed improved perioperative clinical outcome, with significantly higher Trifecta achievement rate, before and after propensity-score matching, with lower perioperative transfusion rate and shorter LOS.	²²
Image guidance improved surgical outcomes	The use of image guidance improved surgical outcomes	They [Campi et al.] concluded that 3DVM assistance allows optimization of postoperative outcomes by ensuring that proper surgical experience and careful preoperative planning are in place and tailoring surgical strategies and techniques according to the individual patient's anatomy.	³¹
Confidence in system	Expressed approval/confidence in image guidance	3DVMs represent a useful tool to plan a tailored surgical approach in case of surgically complex masses.	¹⁵
		3DVT provided precise information of anatomical structure in the operative area and reliable guidance for preoperative plan design.	¹³
		The platform received good feedback from the operating surgeon in all instances with the surgeon commenting particularly on the improved appreciation of hilar vascular anatomy.	²¹
Enhanced surgeon ability/accuracy	Image guidance made the surgeon more accurate and/or improved their operating ability (as stated by the surgeon OR researcher)	The further application of computer-aided technology, based on original image data, has demonstrated to be one of the most effective solutions to supply reliable assistance to surgeons, although shortening the formidable learning curve required for these procedures.	²⁴
Enhanced surgeon ability/accuracy		Overall, it was confirmed that 3DVT enabled surgeons to improve their spatial sensation of the collecting system.	¹³
Anatomical explanation for qualitative assessment	Description of anatomy that points to the need for qualitative assessment	When performing LPN for renal mass, especially complex tumor, complicated region situation may cause some technical challenges for the surgeon, such as invisible feeding arteries and unclear tumor penetration depth.	¹³
Anatomical explanation for qualitative assessment		After exposure of the kidney surface, the 3D model was superimposed over the real anatomy using AR technology to visualize the tumor location, hidden inside the central portion of the parenchyma.	³¹
Subjective statements	Expert opinion without specific data or reference as support	By fusion of the real-time two-dimensional fluoroscopy image with the intraoperative CT images, the surgeon was enabled to understand spatial relations between instruments' tips and tumor margins	²⁶
Subjective statements		It could provide the surgeon with useful assistance and help him achieve better quality of operation.	²²

3DVM = three-dimensional virtual model; CT = computed tomography; IGRAPN = image-guided robot-assisted partial nephrectomy; LOS = length of hospital stay; LPN = laparoscopic partial nephrectomy; OR = odds ratio.

3D-IGS is necessary

This theme is consistent throughout the articles, since proving the necessity of 3D-IGS motivates research to develop these systems. For example, one study states, “Using a three-dimensional virtual imaging of renal tumors instead of 2D CT images with multiplanar reconstructions can help to appreciate anatomical complexity of renal masses.”²² Furthermore, many authors agree that MIPN can be technically challenging owing to variable anatomic complexity. As an example, another study states, “Especially in endophytic tumors or high R.E.N.A.L. Score patients, the accurate understanding of risk structures (bowel, vessels) in relation to target structures is essential.”²⁶

Other researchers report that their “data confirmed again that preoperative planning based on 3D models allows for a better understanding of vascular anatomy.”²⁵ The same group says in a separate study that 3D-IGS systems “[avoid] the building in mind process that the surgeon should do to perceive the features of the organs in a real human body.”¹⁵ From a different perspective, one study cites the loss of feedback in RAPN as the motivation for 3D-IGS saying, “the search for a solution to the loss of haptic feedback has given rise to increasing research efforts in the field of intraoperative image guidance.”²¹

3D-IGS improved surgical outcomes

Another theme that consistently appeared was the assertion that 3D-IGS improved surgical outcomes. Authors often used clinical data and other qualitative statements to support this claim. Authors state their technology “can be useful for identifying the intraparenchymal structures that are difficult to visualize with ultrasound only. This translates to a potential improvement in the quality of the resection phase and a slight reduction of postoperative complications, with better functional recovery.”²⁵ In more specific terms, the same group notes that 3D guidance does “reduce ischemia, positive surgical margins, eGFR, and complications.”¹⁵ Other authors argue, “three-dimensional kidney models for preoperative planning as well as intraoperative guidance could help optimize RAPN perioperative clinical outcomes.”²² Similarly, another study reports “the understanding of the correct volumetric intrarenal tumor expansion, in relation to the instrument position during resection, is an advantage that might reduce positive margin rates, while sparing more nephrons.”²⁶

In addition to improving overall outcomes, preoperative planning with 3D-IGS enabled surgeons to plan for and potentially avoid intraoperative complications.^13,21 Researchers found their systems improved procedural outcomes compared to the control groups (without 3D-IGS, but according to standard of care).

Confidence in system

Overall confidence in the 3D-IGS system, whether it be from the researching authors or the physicians using the system, was identified as essential in determining 3D-IGS's effectiveness. Researchers noted “[the study results] practically proves the clinical role of such new three-dimensional virtual models also in this setting, with perioperative and functional advantages. With these results, their introduction and diffusion in the daily practice are well justified.”¹⁵ In another study, the authors noted, “the increase in the surgeon's self-confidence derived from a perfect preoperative understanding of the patient's regional anatomy resulting from in-depth surgical planning and virtual simulation may be the most important reason for these favorable outcomes.”²⁴ 3D-IGS systems were credited for “[making] the surgeon more confident for tumor excision,”²⁰ and “[enhancing the] surgeon's confidence of surgical procedure and [decreasing] surgical risk and incidence of complications.”¹³

Enhanced surgeon ability

Increasing a surgeon's ability to perform MIPN was identified to be the main underlying motivation for creating 3D-IGS systems. Although the term “ability” is loosely defined in the literature, we use it to encompass notions of surgeon skill and accuracy. Authors stated their system “could provide the surgeon with useful assistance and help him achieve better quality of surgery.”²² In a different study, authors said the platform they were testing, “should result in a more accurate resection phase, guided step by step by augmented reality images,” and “makes it possible to correctly identify the lesion and intraparenchymal structures with a more accurate 3D perception of the location and the nature of the different structures relative to the standard 2D U.S. guidance.”²⁵

Recurring qualitative assessments of increased surgeon accuracy are common throughout the articles in our systematic review. Other authors state their system “improves the surgeon's ability to localize the intrarenal tumors that are not visible on the kidney surface.”²⁰ In the same study, they assert that, “the augmented reality of 3D image fusion on the intraoperative images helps improve surgical accuracy.” Another group says that their system “[allows] for more informed intraoperative decision making.”²¹ Furthermore, other authors stated their platform “facilitates surgeons to obtain an unparalleled guidance and understanding for optimization of surgical strategy.”¹³

Anatomical explanation for qualitative assessment

The articles in our systematic review often describe complex anatomy, which emerged as a theme to justify the common qualitative 3D-IGS assessment. To assess surgeon accuracy improvement, ground truth data are needed for comparison of surgeon performance. This is technically challenging as tumors are often at least partially subsurface or covered in fat. Furthermore, the kidney lacks distinct anatomical landmarks, which can be used as ground truth for accuracy assessment. Thus, the collection of quantitative intraoperative data is made challenging by the complicated anatomy.

As stated in one study, “The management of intrarenal tumors presents a technical challenge during laparoscopic partial nephrectomy because these tumors cannot be visualized on the kidney surface and external visual cues to guide tumor localization or margins are lacking.”²⁰ Another study points out the additional challenge of localizing the vasculature, “when performing laparoscopic partial nephrectomy for renal mass, especially complex tumors, complicated region situation may cause some technical challenges for the surgeon, such as invisible feeding arteries and unclear tumor penetration depth.”¹³ In addition, 3D-IGS could most benefit “patients with complex or endophytic tumors,” but it is these cases in which collecting ground truth tumor location is very challenging.²⁶

Subjective statements

The quotes included in Table 3 indicate instances in which the authors of included studies made claims regarding system efficacy, without specific data or references in support. The comments are likely intended to convey expert opinion rather than quantitative evaluations. The statements are typically in favor of 3D-IGS and support its efficacy and accuracy.

Discussion

This article presents a systematic review of the current literature regarding 3D-IGS implementations for partial nephrectomy. The studies incorporated in this systematic review exhibit a large degree of heterogeneity, reflecting the diverse nature of research in the field of 3D Image Guidance Systems (3D-IGS). This diversity spans not only the study subjects—ranging from phantom experiments and preoperative imaging to animal models and live human subjects—but also the scope of reported results, which include some combination of clinical, nonclinical, and qualitative data. Overall, the literature suggests 3D-IGS improves surgical outcomes by lowering the rate of collecting system violations, risk of surgical complications, eGFR variation, and the loss of renal function. In addition, qualitative analysis highlighted a perceived improvement in surgeon performance and confidence in the 3D-IGS system. Despite these findings, there is no quantitative data regarding surgeon skill improvement within the IGS literature.

Although only eight (32%) studies presented clinical results such as median operation time, median warm ischemia time, final pathology, and excision margins, clinical outcomes point to an increased rate of enucleation and “trifecta” achievement (negative cancer margin, minimal renal functional decrease, and no urological complication). Even in challenging clinical cases, 3D-IGS enabled surgeons to plan and complete the procedures. Although clinical data are promising, the eligible studies include 793 nonrandomized patients. Large-scale studies will need to take place across various surgical centers before conclusions can be drawn solely with clinical data.

Registration is a required component of 3D-IGS systems, which intend to inform intraoperative decisions. The model must be oriented in an accurate representation of the operative field. This review revealed manual registration to be the most common method of registration as the optimal computed registration method has not been determined for partial nephrectomy. Although manual registration is the simplest to implement, it requires constant input from a technician or the surgeon. Furthermore, systems utilizing manual registration can only offer a certain level of benefit beyond the standard of care because manual (or, semimanual) registration has much higher error than fully computed.³²

Fewer studies implemented a computed registration, likely attributable to the technical challenges that have yet to be overcome. Since the kidney surface offers no distinct anatomical landmark, fiducials must be inserted to employ point-based registration.^23,26 To avoid the morbidity of inserting fiducials, surface-based registration has been implemented, but requires accurate surface scanning data and/or the application of ink fiducials. Work is ongoing to find a computed registration method that minimizes invasiveness and maximizes accuracy. Among the studies in our systematic review, both registration methods have proven to be valid for partial nephrectomy. Manual registration has a smaller technical barrier but requires constant input with lower accuracy and no performance guarantee. Computed registration, in contrast, has the potential to offer better accuracy, but currently requires more time for registration data collection, and requires further research before a consensus is reached on the preferred method of data collection and registration for MIPN.

Registration limitations were the most common challenge noted across the studies we reviewed, with soft tissue deformation being the most cited issue. It is an active area of research in the engineering community and many promising techniques are under development.⁵⁰ Many other factors such as cost, limited sample size, and the inclusion of highly experienced surgeons were also listed as limitations.

It is important to note that, while one group stated the split view between the 3D-IGS (TilePro for the da Vinci) and the endoscope as a limitation, other studies utilizing the same display method did not. And indeed, some have intentionally designed their system in this way, since it reduces visual clutter in endoscope images, and removes endoscope to tissue registration as a source of error in the overall IGS system. (Displays that do not attempt to fuse IGS with endoscope images do not need to know exactly where the endoscope is to be accurate in their displaying tool locations relative to anatomy). Among the studies in our systematic review, it is recognized that the reconstruction of a 3D model from 2D data is not without challenges.

Segmenting image sets manually can take many hours (10+) to do carefully and thoroughly and is prone to human error even when segmented by experts. Automated computational segmentation methods have been developed to create the model before and during the procedure but remain an active area of research. The segmentation method used in prior studies varies, and whether it has an overall influence on outcomes is currently unknown, with no study to date yet specifically investigating this. Finally, on a different topic, one study cited unclear significance of results as a key limitation.⁴⁸ This study required manual definition of tumor location on the segmented kidney during preoperative planning, using information from 2D scans. The tumor placement error, although small, could point to the necessity of 3D-IGS.

Among the studies that employed in vivo registration, clinical outcomes and qualitative statements regarding surgeon accuracy/ability improvement were reported as primary results. Although surgeon accuracy improvement appeared as a common theme across this subset of studies, it has not been measured quantitatively. The absence of quantitative assessment in the clinical studies motivated the qualitative analysis we conducted in this article.

The qualitative analysis serves to better understand the underlying themes noted in the literature. It is useful to draw conclusions on the overall efficacy of 3D-IGS for partial nephrectomy as it stands today. The results suggest that the researchers developing and testing 3D-IGS systems believe them to be an effective tool for increasing surgeon confidence and skill. They believe its integration into the standard of care is highly motivated, and that 3D-IGS has the potential to make complex partial nephrectomy more accessible to less experienced surgeons. Furthermore, they argue that its use can improve surgical outcomes. When used for preoperative planning, they note that it can enable surgeons to recognize potential intraoperative risks and enact plans to minimize them. There is also a noteworthy absence of quantitative clinical accuracy data, since establishing ground truth is challenging in a clinical setting.

Overall, the literature suggests a promising future for 3D-IGS. One potential advancement of this technology could involve the refinement and validation of automated organ deformation correction methods, incorporating not only machine learning but also real-time feedback from intraoperative imaging. This advancement aims to enhance the accuracy and adaptability of 3D-IGS as soft tissues deform during operation. In addition, the integration of artificial intelligence (AI) and machine learning with 3D-IGS systems holds the potential to revolutionize operation through autonomous robotic procedures.

Future research might focus on the development of AI-driven surgical systems that can collaborate with surgeons, offering real-time decision support, enhancing precision, and further reducing invasiveness. As the systems described in this article advance, they could expand their clinical utility to encompass a wider range of soft-tissue-sparing operations. Beyond nephron-sparing procedures, 3D-IGS has potential applications in surgeries such as lung segmentectomies, lumpectomies, and partial prostatectomies. These potential research directions and clinical applications underscore the transformative impact that 3D-IGS could have on the field of minimally invasive surgery.

Conclusion

This review synthesizes both clinical and nonclinical data to present the latest information regarding 3D-IGS's effectiveness for MIPN. 3D-IGS research is ongoing, but current data suggests it is a useful tool to have in the operating room. The studies in our systematic review employed various 3D-IGS platforms in a variety of study designs. Model-to-operative field registration remains the most challenging aspect of these platforms. Furthermore, quantitative data regarding surgeon skill improvement are lacking in the current literature. Clinical outcomes reported in the literature we reviewed are promising, but require larger scale, randomized trials to be conclusive. Thus, the benefit of 3D-IGS is becoming clear from the consensus in the literature, but more research is required before it can be widely clinically adopted.

Authors' Contribution

Conception and design of the study were contributed by S.H., R.W., and N.K. Acquisition of data was done by P.C. and S.S. Analysis and interpretation were carried out by P.C., S.S., and S.K.G. Drafting the article was done by P.C. Revision of the article was done by P.C. and S.S. Supervision was done by S.H., R.W., and N.K.

Footnotes

Author Disclosure Statement

No competing financial interests exist.

Funding Information

This work was supported, in part, by NIH-NIBIB training grant T32-EB021937 and R01 EB023717.

Abbreviations Used

References

Suk-Ouichai

, Tanaka

, Wang

, et al. Renal cancer surgery in patients without preexisting chronic kidney disease—Is there a survival benefit for partial nephrectomy?. J Urol, 2019; 201(6):1088–1096; doi: 10.1097/JU.0000000000000060

Jaffray

Minimally invasive surgery. Arch Dis Child, 2005; 90(5):537–542; doi: 10.1136/adc.2004.062760

Robinson

, Stiegmann

. Minimally invasive surgery. Endoscopy, 2004; 36(1):48–51; doi: 10.1055/s-2004-814113

Malthouse

, Kasivisvanathan

, Raison

, et al. The future of partial nephrectomy. Int J Surg, 2016; 36:560–567; doi: 10.1016/j.ijsu.2016.03.024

Gadus

, Kocarek

, Chmelik

, et al. Robotic partial nephrectomy with indocyanine green fluorescence navigation. Contrast Media Mol Imaging, 2020; 2020:1–8; doi: 10.1155/2020/1287530

Zhang

, Ouyang

, Wu

, et al. Robot-assisted partial nephrectomy with a standard laparoscopic ultrasound probe in treating endophytic renal tumor. Asian J Surg, 2020; 43(2):423–427; doi: 10.1016/j.asjsur.2019.07.005

Alenezi

, Karim

. Role of intra-operative contrast-enhanced ultrasound (CEUS) in robotic-assisted nephron-sparing surgery. J Robot Surg, 2015; 9(1):1–10; doi: 10.1007/s11701-015-0496-1

Fotiadis

, Sabharwal

, Gangi

, et al. Combined MRI and fluoroscopic guided radiofrequency ablation of a renal tumor. Cardiovasc Intervent Radiol, 2009; 32(1):184–187; doi: 10.1007/s00270-008-9334-4

Miki

, Shimomura

, Yamada

, et al. Percutaneous cryoablation of renal cell carcinoma guided by horizontal open magnetic resonance imaging. Int J Urol, 2006; 13(7):880–884; doi: 10.1111/j.1442-2042.2006.01432.x

10.

Balachandran

, Schurzig

, Fitzpatrick

, et al. Evaluation of portable CT scanners for otologic image-guided surgery. Int J Comput Assist Radiol Surg, 2012; 7(2):315–321; doi: 10.1007/s11548-011-0639-4

11.

Nimmagadda

, Ferguson

, Kavoussi

, et al. Patient-specific, touch-based registration during robotic, image-guided partial nephrectomy. World J Urol, 2022; 40(3):671–677; doi: 10.1007/s00345-021-03745-y

12.

L-M

, Vagvolgyi

, Agarwal

, et al. Augmented reality during robot-assisted laparoscopic partial nephrectomy: Toward real-time 3D-CT to stereoscopic video registration. Urology, 2009; 73(4):896–900; doi: 10.1016/j.urology.2008.11.040

13.

Wang

, Qi

, Yuan

, et al. Application of three-dimensional visualization technology in laparoscopic partial nephrectomy of renal tumor: A comparative study. J Laparoendosc Adv Surg Tech, 2017; 27(5):516–523; doi: 10.1089/lap.2016.0645

14.

Galloway

RL.

The process and development of image-guided procedures. Ann Rev Biomed Eng, 2001; 3(1):83–108; doi: 10.1146/annurev.bioeng.3.1.83

15.

Amparore

, Pecoraro

, Piramide

, et al. Three-dimensional imaging reconstruction of the kidney's anatomy for a tailored minimally invasive partial nephrectomy: A pilot study. Asian J Urol, 2022; 9(3):263–271; doi: 10.1016/j.ajur.2022.06.003

16.

Hughes-Hallett

, Pratt

, Mayer

, et al. Using preoperative imaging for intraoperative guidance: A case of mistaken identity. Int J Med Robot Comp Assist Surg, 2016; 12(2):262–267; doi: 10.1002/rcs.1654

17.

, Yang

, et al. Meta grayscale adaptive network for 3D integrated renal structures segmentation. Med Image Anal, 2021; 71:102055; doi: 10.1016/j.media.2021.102055

18.

Wang

, Suenaga

, Liao

, et al. Real-time computer-generated integral imaging and 3D image calibration for augmented reality surgical navigation. Comput Med Imag Graphics, 2015; 40:147–159; doi: 10.1016/j.compmedimag.2014.11.003

19.

Schneider

, Nguan

, Longpre

, et al. Motion of the kidney between preoperative and intraoperative positioning. IEEE Trans Biomed Eng, 2013; 60(6):1619–1627; doi: 10.1109/TBME.2013.2239644

20.

Chen

, Li

, Wu

, et al. Surgical planning and manual image fusion based on 3D model facilitate laparoscopic partial nephrectomy for intrarenal tumors. World J Urol, 2014; 32(6):1493–1499; doi: 10.1007/s00345-013-1222-0

21.

Hughes-Hallett

, Pratt

, Mayer

, et al. Image guidance for all—TilePro display of 3-dimensionally reconstructed images in robotic partial nephrectomy. Urology, 2014; 84(1):237–243; doi: 10.1016/j.urology.2014.02.051

22.

Michiels

, Khene

Z-E

, Prudhomme

, et al. 3D-Image guided robotic-assisted partial nephrectomy: A multi-institutional propensity score-matched analysis (UroCCR study 51). World J Urol, 2023; 41(2):303–313; doi: 10.1007/s00345-021-03645-1

23.

Teber

, Guven

, Simpfendörfer

, et al. Augmented reality: A new tool to improve surgical accuracy during laparoscopic partial nephrectomy? Preliminary in vitro and in vivo results. Eur Urol, 2009; 56(2):332–338; doi: 10.1016/j.eururo.2009.05.017

24.

Wang

, Zhang

, Yuan

, et al. Preoperative planning and real-time assisted navigation by three-dimensional individual digital model in partial nephrectomy with three-dimensional laparoscopic system. Int J Comput Assist Radiol Surg, 2015; 10(9):1461–1468; doi: 10.1007/s11548-015-1148-7

25.

Porpiglia

, Checcucci

, Amparore

, et al. Three-dimensional augmented reality robot-assisted partial nephrectomy in case of complex tumours (PADUA ≥10): A new intraoperative tool overcoming the ultrasound guidance. Eur Urol, 2020; 78(2):229–238; doi: 10.1016/j.eururo.2019.11.024

26.

Simpfendörfer

, Gasch

, Hatiboglu

, et al. Intraoperative computed tomography imaging for navigated laparoscopic renal surgery: First clinical experience. J Endourol, 2016; 30(10):1105–1111; doi: 10.1089/end.2016.0385

27.

Ong

, Glisson

, Altamar

, et al. Intraprocedural registration for image-guided kidney surgery. IEEE/ASME Transact Mech, 2010; 15(6):847–852; doi: 10.1109/TMECH.2010.2066985

28.

Burgner

, Simpson

, Fitzpatrick

, et al. A study on the theoretical and practical accuracy of conoscopic holography-based surface measurements: toward image registration in minimally invasive surgery. Int J Med Robot Comput Assist Surg, 2013; 9(2):190–203; doi: 10.1002/rcs.1446

29.

Nosrati

, Abugharbieh

, Peyrat

J-M

, et al. Simultaneous multi-structure segmentation and 3D nonrigid pose estimation in image-guided robotic surgery. IEEE Trans Med Imaging, 2016; 35(1):1–12; doi: 10.1109/TMI.2015.2452907

30.

Altamar

, Ong

, Glisson

, et al. Kidney deformation and intraprocedural registration: A study of elements of image-guided kidney surgery. J Endourol, 2011; 25(3):511–517; doi: 10.1089/end.2010.0249

31.

Amparore

, Piramide

, Pecoraro

, et al. Identification of recurrent anatomical clusters using three-dimensional virtual models for complex renal tumors with an imperative indication for nephron-sparing surgery: New technological tools for driving decision-making. Eur Urol Open Sci, 2022; 38:60–66; doi: 10.1016/j.euros.2022.02.006

32.

Pratt

, Mayer

, Vale

, et al. An effective visualisation and registration system for image-guided robotic partial nephrectomy. J Robot Surg, 2012; 6(1):23–31; doi: 10.1007/s11701-011-0334-z

33.

Kokko

, Van Citters

, Seigne

, et al. A particle filter approach to dynamic kidney pose estimation in robotic surgical exposure. Int J Comput Assist Radiol Surg, 2022; 17(6):1079–1089; doi: 10.1007/s11548-022-02638-8

34.

Kokko

, Seigne

, Van Citters

, et al. Modeling the Surgical Exposure of Anatomy in Robot-Assisted Laparoscopic Partial Nephrectomy. In: Medical Imaging 2020: Image-Guided Procedures, Robotic Interventions, and Modeling. (Fei B, Linte CA. eds). SPIE: Houston, Texas United States; 2020; p. 52; doi: 10.1117/12.2550605

35.

Baumhauer

, Feuerstein

, Meinzer

H-P

, et al. Navigation in endoscopic soft tissue surgery: Perspectives and limitations. J Endourol, 2008; 22(4):751–766; doi: 10.1089/end.2007.9827

36.

Friets

, Bieszczad

, Kynor

, et al. Endoscopic Laser Range Scanner for Minimally Invasive, Image Guided Kidney Surgery. In: Medical Imaging 2013: Image-Guided Procedures Robotic Interventions, and Modeling (Holmes DR, Yaniv ZR. eds). SPIE: Lake Buena vista, Florida, 2013; p. 867105; doi: 10.1117/12.2007608

37.

Siebold

, Ferguson

, Pitt

, et al. Choosing Statistically Safe, Variable-Thickness Margins in Robot-Assisted Partial Nephrectomy. In: 2020 International Symposium on Medical Robotics (ISMR) IEEE. IEEE: Atlanta, GA, USA, 2020; pp. 152–158; doi: 10.1109/ISMR48331.2020.9312947

38.

Edgcumbe

, Singla

, Pratt

, et al. Follow the light: projector-based augmented reality intracorporeal system for laparoscopic surgery. J Med Imag, 2018; 5(02):1; doi: 10.1117/1.JMI.5.2.021216

39.

Chauvet

, Collins

, Debize

, et al. Augmented reality in a tumor resection model. Surg Endosc, 2018; 32(3):1192–1201; doi: 10.1007/s00464-017-5791-7

40.

Kong

S-H

, Haouchine

, Soares

, et al. Robust augmented reality registration method for localization of solid organs' tumors using CT-derived virtual biomechanical model and fluorescent fiducials. Surg Endosc, 2017; 31(7):2863–2871; doi: 10.1007/s00464-016-5297-8

41.

Zhang

, Wang

, et al. A markerless automatic deformable registration framework for augmented reality navigation of laparoscopy partial nephrectomy. Int J Comput Assist Radiol Surg, 2019; 14(8):1285–1294; doi: 10.1007/s11548-019-01974-6

42.

Dixon

, Daly

, Chan

, et al. Surgeons blinded by enhanced navigation: The effect of augmented reality on attention. Surg Endosc, 2013; 27(2):454–461; doi: 10.1007/s00464-012-2457-3

43.

Popay

, Roberts

, Sowden

, et al. Guidance on the conduct of narrative synthesis in systematic reviews. ESRC Methods Programme, 2006; 15:47–71.

44.

Socio Cultural Research Consultants, LLC, Dedoose Version 9.0.17. Cloud Application for Managing, analyzing, and presenting quantitative and mixed method research data. SocioCultural Research Consultants, LLC: Los Angeles, CA, 2021. Available from: www.dedoose.com

45.

Saldaña

Fundamentals of Qualitative Research. Understanding Qualitative Research. Oxford University Press: New York, NY; 2011.

46.

Hong

, Fàbregues

, Bartlett

, et al. The Mixed Methods Appraisal Tool (MMAT) version 2018 for information professionals and researchers. Educ Inform, 2018; 34(4):285–291; doi: 10.3233/EFI-180221

47.

Baumhauer

, Simpfendörfer

, Müller-Stich

, et al. Soft tissue navigation for laparoscopic partial nephrectomy. Int J Comput Assist Radiol Surg, 2008; 3(3–4):307–314; doi: 10.1007/s11548-008-0216-7

48.

Wake

, Wysock

, Bjurlin

, et al. “Pin the Tumor on the Kidney”: An evaluation of how surgeons translate CT and MRI data to 3D models. Urology, 2019; 131:255–261; doi: 10.1016/j.urology.2019.06.016

49.

Hung

, Cai

, Simmons

, et al. “Trifecta” in partial nephrectomy. J Urol, 2013; 189(1):36–42; doi: 10.1016/j.juro.2012.09.042

50.

Markelj

, Tomaževič

, Likar

, et al. A review of 3D/2D registration methods for image-guided interventions. Med Image Anal, 2012; 16(3):642–661; doi: 10.1016/j.media.2010.03.005