Microsoft HoloLens 2 vs. Tablet-based Augmented Reality and 3D printing for fronto- orbital reconstruction of craniosynostosis: A case study

doi:10.21203/rs.3.rs-5097210/v1

Download PDF

Research Article

Microsoft HoloLens 2 vs. Tablet-based Augmented Reality and 3D printing for fronto- orbital reconstruction of craniosynostosis: A case study

https://doi.org/10.21203/rs.3.rs-5097210/v1

This work is licensed under a CC BY 4.0 License

Journal Publication

published 21 Mar, 2025

Read the published version in 3D Printing in Medicine →

You are reading this latest preprint version

Background

Craniosynostosis is a congenital condition characterized by the premature fusion of cranial sutures, leading to potential complications such as abnormal skull growth, increased intracranial pressure, and cognitive delays. Traditionally, open cranial vault reconstruction (OCVR) has been used to treat this condition. However, it is highly subjective and greatly dependent on the surgeon's expertise, which can lead to residual deformities and the need for reoperation. Effective preoperative planning can greatly improve surgical outcomes, although the major challenge is accurately translating this plan into the clinical setting. Recently, augmented reality (AR) and 3D printing have emerged as promising technologies to facilitate this endeavor. In this work, we propose three alternatives, leveraging these technologies, to guide the precise repositioning of remodeled bone fragments in the patient.

Methods

The three guidance methods are AR on a tablet, AR with Microsoft HoloLens 2, and 3D-printed spacers. The accuracy of each method was assessed by measuring the deviation of each bone fragment from the virtual surgical plan (VSP) in a simulated environment using 3D-printed phantoms based on a 14-month-old boy with trigonocephaly. The same assessment was also performed during his actual surgery.

Results

All three guidance methods demonstrated similar levels of accuracy, with mean placement errors below 1 mm in all cases. The AR systems allowed for real-time adjustments, enhancing precision. Statistical analysis showed no significant differences in error rates between the different methods or attempts.

Conclusions

Integrating AR and 3D printing into craniosynostosis surgery holds great potential for improving OCVR. While 3D-printed spacers are useful when digital technologies are unavailable, AR-based methods provide more comprehensive guidance. Nevertheless, our study suggests that the choice may depend more on the specific clinical context, user-specific skills, and available resources rather than on a clear superiority of one method over the others.

Augmented reality

Craniosynostosis

Microsoft HoloLens 2

Tracking

Surgical guidance

Computer-assisted interventions

Craniosynostosis is a congenital condition characterized by the premature fusion of one or more cranial sutures. This can lead to abnormal skull and restricted brain growth perpendicular to the affected suture [1]. The incidence of this condition ranges from 1 in 2000 to 1 in 2500 live births and can occur as part of a syndrome or as an isolated defect (nonsyndromic) [2]. If untreated, craniosynostosis can cause increased intracranial pressure, impaired brain growth, visual problems, and cognitive delay [3].

Among the various treatment options, open cranial vault reconstruction (OCVR) remains the most common method. These surgeries involve removing the affected bones, reshaping them to the most suitable form for the patient, and then securing them back in position. On the other hand, minimally invasive treatments, such as spring-mediated cranioplasty and endoscopically assisted craniosynostosis surgery (EACS) combined with helmet therapy, are slowly gaining popularity [4], [5]. Several works have demonstrated that both types of treatments have similar complication rates, and there is not a clear superiority of one approach over the other[6], [7], [8]. EACS offers reduced surgical incision size, lower reoperation rates, and shorter hospital stays [9]. However, OCVR is still the preferred alternative when immediate correction of skull shape and intracranial pressure is required. Moreover, it permits a more targeted surgical intervention with customized virtual surgical planning and a potential reduction in persistent bony defects [10]. In general terms, most studies agree that the ideal time to perform early EAS is within the child’s first 3 months of life since the skull is soft, with a thickness < 2 mm in the frontal and parietal bones facilitating strip craniectomies. In turn, OCVR is more convenient when the patient exceeds the nine months of life [9].

Regardless of the treatment chosen, diagnosing and surgically correcting craniosynostosis largely relies on the surgeon's subjective assessment. The surgeon typically determines the severity of the cranial deformity and develops a strategy for reshaping the affected bone to restore a normal skull shape. This approach relies heavily on the surgeon's skill and experience, leading to high variability in surgical outcomes and highlighting the need for advanced techniques to improve accuracy and reproducibility.

Effective preoperative planning is essential for achieving optimal surgical outcomes in craniosynostosis treatment. As such, previous research groups have worked on developing adequate surgical plans based on normative cranial shape models [11], [12]. Employing these customized models can enhance surgical outcomes and increase the objectivity and reproducibility of procedures [13], [14]. However, the main challenge lies in accurately translating this plan into the clinical setting, as any inaccuracies in bone placement can lead to residual deformities. This issue has prompted the exploration of advanced technologies to help bridge the gap between planning and execution. Three-dimensional (3D) technologies have shown great promise in this regard. For instance, 3D printing has been used to create patient-specific models and surgical guides, allowing surgeons to better translate the surgical plan into clinical practice [15], [16]. On the other hand, surgical navigation systems, known for their high precision, have also been employed to improve the accuracy of cranial remodeling procedure [17], [18]. In previous work, we presented and evaluated a novel workflow for accurately translating the virtual surgical plan (VSP) into the operating room (OR) using both 3D printing and optical navigation [19]. However, traditional navigation systems are often bulky, time-consuming to set up, and can be counterintuitive as they display information on external screens, diverting the surgeon's focus away from the patient.

Augmented Reality (AR) has emerged as a promising alternative in recent years, offering the ability to overlay virtual information directly onto the surgical field so that surgeons maintain focus on the patient. AR is more compact and affordable than traditional navigation systems and can be used with smartphones/tablets or head-mounted displays (HMDs). Despite its potential, AR has seen limited use in craniosynostosis surgeries, with most applications restricted to simulation scenarios [20], [21]. When intraoperative results are available, they are typically based on subjective feedback from surgeons [22] or assessed using broad metrics such as changes in intracranial volume before and after surgery [23]. These studies often lack precise measurements that directly evaluate the accuracy of bone positioning using AR techniques.

In this work, we developed, analyzed, and compared alternative 3D solutions to improve the placement accuracy of remodeled bones in OCVR surgeries. Two of these solutions involved AR applications: one designed for an Android tablet and the other for Microsoft HoloLens 2. The third method was based on the use of 3D-printed references. We initially tested these methods in a simulated environment using a 3D-printed patient-specific phantom and subsequently applied them during an actual surgical procedure. Our primary objective was to evaluate the accuracy of bone placement achieved with each of these guidance methods individually. We focused specifically on the deviation of each positioned bone from the VSP to identify the most effective approach for clinical use.

2.1. Clinical data

The proposed solutions were evaluated on a 14-month-old boy diagnosed with trigonocephaly, treated at the Hospital General Universitario Gregorio Marañón (HGUGM) in Madrid, Spain. The standard surgical procedure for this type of craniosynostosis (metopic) involves the remodeling of the frontal bone and the supraorbital bar (hereafter referred to as the SO bar) [24]. These are the bone fragments we focus on in this work. The patient was nonsyndromic, had no previous craniofacial surgeries, and did not exhibit hydrocephalus, intracranial hemorrhages, or craniofacial trauma.

2.2. Virtual surgical planning and surgical guide design

A preoperative computed tomography (CT) scan of the patient’s head was acquired using a Philips Mx8000 CT scanner with a slice thickness of 0.625 mm. A 3D virtual model of the patient’s skull was generated from these CT images using the Segment Editor module in 3D Slicer [25]. A VSP was outlined to achieve the ideal skull shape for this patient using normative cranial shape models. Specifically, we used the statistical atlas generated at Children’s National Hospital from CT images of 100 healthy infants to develop an individualized VSP based on the patient’s age and gender [11], [12]. The VSP outlines how the bones should be cut, reshaped, and repositioned to minimize cranial shape abnormalities and achieve the desired outcome.

Following the methodology from our previous work [19], 3D-printed patient-specific surgical guides were employed for cranial osteotomies and bone remodeling. The cutting guides, created from the patient’s skull shape extracted from the preoperative CT scan, provided precise cutting lines for the osteotomy of the frontal bone and SO bar. Remodeling templates, derived from the VSP, served as molds to reshape the bone fragments to the desired form.

To place the remodeled bones back in the patient in the expected positions, this work presents and analyzes three alternative methods: 3D-printed spacers, AR guidance with a tablet, and AR guidance with Microsoft HoloLens 2. Each method also utilized patient-specific 3D-printed surgical guides. The 3D-printed spacers were designed to fit into the SO bar gap after the osteotomy and physically constrain the advance of the remodeled SO bar. However, they do not provide any guidance for the placement of the frontal bone. For the AR methods, AR guides were created following the same protocol as with the cutting guides, but including the design of an adapter to hold the AR markers that enable automatic registration of virtual information. All designs were exported in STL format for 3D printing and subsequently converted to OBJ format to develop the AR applications for Microsoft HoloLens 2 and tablet in Unity.

2.3. Surgical workflow

After defining the individualized VSP from normative cranial shape models, the proposed surgical workflow consists of ten steps (Fig. 1). The procedure starts with bicoronal S-shaped incision to expose the frontal and supraorbital bone regions. Secondly, the surgeon fits the 3D-printed cutting guides onto the exposed bones and uses an ink-compatible marker to mark the osteotomy lines. These guides also serve as a reference to fit the AR guides in the patient’s left and right parietal bones. The AR guides are affixed to the skull by creating 2 mm diameter holes in the bone tissue with three resorbable pins (SonicPins Rx, KLS Martin, Germany) in each case. After fixing the AR guides, the cutting guides are removed, and osteotomies are performed. The frontal bone and the SO bar are then reshaped using remodeling templates and positioned back in the head using three distinct methods: 3D printed spacers (only for the SO bar) and AR guidance with tablet or Microsoft HoloLens 2. To finish, the remodeled bones are secured to the surrounding healthy bones with resorbable plates and screws. The AR guides are then removed, and a final scan of the patient is acquired to verify the result before closing the incision.

2.4. Augmented reality reference markers and applications

We developed two AR applications for surgical guidance using Unity 2022.3.20f1 and C# programming language. The first one is compatible with any Android device (Fig. 2A). Following previous works of our group for surgical guidance in cranial [22] and maxillofacial reconstruction [26], the application displays the ideal position of the remodeled bones in the patient's head. Automatic registration between the virtual and real worlds is achieved using the Vuforia software development kit (SDK), version 10.22.5 (Parametric Technology Corporation Inc., Boston, MA, USA). It detects two planar AR reference markers in the camera’s field of view (FOV) which contain specific black and white patterns in two-dimensional (2D) 40 x 40 mm squares. Both references have a five-star rating in the Vuforia API, which is the maximum score for marker quality and indicates that they will be accurately tracked. Given the placement of the AR markers on the patient (one on each side of the head), only one can be detected at a time. Based on the detected marker and its estimated pose, the application renders the virtual models in their designated positions within the surgical site.

In addition to displaying the remodeled bones, the application also shows the virtual position of other surgical guides and a frame around the detected marker for visual verification of proper alignment between the virtual and real worlds. The frame is also color-coded and transitions from green to red to indicate when tracking is lost. In that case, the models remain in their corresponding position where the AR marker was detected last. The display properties of each model can be tailored by selecting them in the dropdown menu (top-left corner of the screen) and then manipulating the visibility toggle and the transparency slider on the screen. The option “All models” applies the modifications to all virtual models simultaneously. The remodeled virtual bone fragments were split into right and left halves. When detecting each marker, only the corresponding half of the bone fragments was displayed to prevent depth perception issues when transparency was increased.

The second application is the direct translation of the Android application to a Head Mounted Display (HMD) environment. Specifically, it was developed for Microsoft HoloLens 2 using Mixed Reality Toolkit (MRTK) version 2.8.3.0. This application contains the same user interface as the Android application, but it is adapted to the MRTK environment (Fig. 2B). Two additional buttons were added as shortcuts to 30% and 100% opacity. In this case, leveraging the AR glasses capabilities, all buttons can be called using voice commands such as “frontal left” to select the left region of the frontal bone, “toggle visibility” to turn it on or off, and “transparent” or “opaque”. The transparency slider must be manually manipulated when a finer tuning of the model’s transparency is desired.

2.5. 3D printing and sterilization

For their use in the surgical scenario, surgical cutting guides and remodeling templates were fabricated through additive manufacturing with selective laser sintering in polyamide material with an external provider (KLS Martin Group, Tuttlingen, Germany). The AR guides and reference markers for the AR applications were fabricated in-house at HGUGM, which is certified as a manufacturer for this purpose [27]. The AR reference markers were 3D printed in black and white polylactic acid (PLA) through the fused deposition modeling (FDM) technique on a double extruder Raise 3D Pro 2 3D printer. The AR guides were fabricated using the stereolithography technique (SLA) with a Formlabs Form 2 3D printer (Formlabs Inc., Somerville, MA, USA) using biocompatible BioMed Clear V1 resin. This resin possesses USP class IV certification, ensuring its suitability for direct contact with the patient’s bone for a longer time [28]. All parts were sterilized before surgery at HGUGM. The cutting guides, remodeling templates, and AR markers were sterilized with water vapor in a single cycle of 75 minutes, with a peak temperature of 134°C for a minimum time of 5 minutes. The AR guides were sterilized using hydrogen peroxide with a peak temperature of 55°C for a minimum of 3 minutes in a single cycle of one hour to prevent deformation.

2.6. Performance evaluation

Before the surgery, we conducted experiments in a simulation scenario using 3D-printed patient-specific phantoms. Five users, including four researchers and the operating surgeon, independently placed the SO bar and the frontal bone phantoms in their corresponding position in the skull phantom sequentially using one of the three guidance methods (AR applications in the tablet or Microsoft HoloLens 2 for both bone fragments and 3D-printed spacers only for the SO bar). In all cases, the bones were affixed to the skull using modeling plaster. The order of the guidance method employed was randomized, and the setup was reset between users, bone fragments, and guidance methods. Each user performed a single attempt with each bone and method, yielding a total of 25 simulations.

During the surgical procedure, the remodeled SO bar fragment was placed three times, each time using one of the three guidance methods (Fig. 3). While the outcomes of these attempts could be successfully recorded for further evaluation, time constraints prevented the same process from being applied to the frontal bone. Consequently, all records in the operating room pertain only to the SO bar.

In all cases, the relative position of the placed bone fragment to the patient's head was recorded using an Artec Eva (Artec3D, Senningerber, Luxembourg) structured light scanner. After each scan, the bone fragments were removed to reset the positioning process before applying the next guidance method. To evaluate the results, we generated 3D models from the scans and registered them to the VSP using 3D Slicer. The global positioning error was calculated from distance maps created with the Model to Model Distance module in 3D Slicer, which measures the distance from each point on the placed bone model to the nearest point on the ideal model in the VSP. The error was also analyzed by axis for a more detailed understanding of the results. Additionally, we measured the advancement error of the SO bar, defined as the anteroposterior distance between the ends of the scanned SO bar and its counterpart in the virtual plan. For a more detailed explanation of the phantom design, surgical experience, and evaluation protocol used to analyze the data recorded, please refer to the Supplementary Document, available at […]. Complementing the Supplementary Document, two supplementary videos are available at the same link. Supplementary Video 1 provides footage from the surgical procedure, offering a complete overview of the experience, while Supplementary Video 2 demonstrates the workflow used to evaluate the data from the surgical scenario.

The mean translation and rotation errors for each bone fragment are presented in Table 1, categorized by scenario and guidance method. The table also presents the mean placement errors derived from the Euclidean distance between the models for each case. Moreover, Fig. 4 illustrates the distance maps generated for the SO bar fragments positioned using each of the three guidance methods during surgery, compared to the ideal plan. The maximum errors observed were 2.0 mm for both the tablet and Microsoft HoloLens 2, and 2.5 mm for the 3D-printed spacers. Nonetheless, the 90th percentile error in each case is 0.9 mm for the tablet, 1.1 mm for Microsoft HoloLens 2, and 0.7 mm for the 3D-printed spacers. All errors were calculated based on absolute values.

Table 1

Mean translation and rotation errors obtained for each bone and scenario, by guidance method. The last column represents the final error obtained by considering the overall positioning. "HL2" stands for Microsoft HoloLens 2, and "Spacers" refers to 3D-printed spacers.
		Mean translation error ± std (mm)			Mean rotation error ± std (°)			Mean placement error ± std (mm)
		Tablet	HL2	Spacers	Tablet	HL2	Spacers	Tablet	HL2	Spacers
SO bar	Surgical scenario	0.6 ± 0.6	0.8 ± 0.3	0.2 ± 0.1	0.8 ± 0.5	0.8 ± 0.4	1.1 ± 0.7	0.5 ± 0.3	0.6 ± 0.3	0.3 ± 0.4
SO bar	Simulation scenario	0.6 ± 0.4	0.6 ± 0.4	0.6 ± 0.4	0.4 ± 0.5	0.6 ± 0.8	0.6 ± 0.6	0.6 ± 0.4	0.6 ± 0.5	0.5 ± 0.4
Frontal bone	Simulation scenario	0.7 ± 0.3	0.6 ± 0.4	-	0.9 ± 0.6	0.9 ± 0.6	-	0.6 ± 0.4	0.6 ± 0.4	-

We performed a series of statistical tests to analyze the recorded data. A Shapiro-Wilk test confirmed that the data for all attempts and guidance methods followed a normal distribution, and the Levene test indicated homogeneous variance across these groups. We called “attempts” the data grouped by users (four researchers and the surgeon in the simulation scenario plus the same surgeon during surgery). Based on this, we conducted several ANOVA tests to compare the error rates among the different attempts and guidance methods. The analysis revealed no significant differences in error between the attempts or among the guidance methods, with a significance level set at 0.05 for all tests.

Figure 5 breaks down the translation and rotation errors by axis, encompassing all data recorded in the simulation scenario and, where applicable, during surgery. Translation along the R, A, and S axes represents the displacement of the positioned bone fragment to the right, anterior, or superior directions relative to the VSP, measured in millimeters. Rotation along the R axis indicates how much the positioned fragment is tilted forward or backward compared to the reference model, measured in degrees. Rotation along the A axis measures the degree of tilt to the right or left, while rotation along the S axis assesses the model's torsion around the patient’s vertical axis.

For both bone fragments, the translation and rotation data grouped by axis met the assumptions of normality and homogeneity of variance according to the Shapiro-Wilk and Levene tests, respectively. For the SO bar, an ANOVA test indicated statistically significant differences between translation axes, and a Tukey HSD test revealed that the significant difference is between the A and S axes. Similarly, significant differences were found in the rotation around the A axis. In the case of the frontal bone, the Tukey HSD test shows significant differences between all pairs of axes for both translation and rotation, with the greatest variability observed along the S axis in both translation and rotation. For all translation and rotation axes, the ANOVA analyses did not detect statistically significant differences when comparing the guidance methods used (Tablet, Microsoft HoloLens 2, and 3D printed spacers).

To further analyze the SO bar placement, we measured its advancement along the antero-posterior axis. Figure 6 shows the distance from each of the left and right ends of the placed SO bars to their corresponding positions in the VSP. Positive and negative distances indicate that the placed SO bar is shifted towards the posterior and anterior directions, respectively. The average advancement error, considering both left and right sides, was 0.8 ± 0.4 mm, with mean errors of 0.7 ± 0.5 mm on the left side and 0.8 ± 0.5 mm on the right side. The maximum error recorded was 2.0 mm on the right side using the tablet application during surgery. Shapiro-Wilk and Levene tests confirmed that the data follow a normal distribution with homogeneous variance when grouped by attempt, guidance method, or side (right / left). Again, ANOVA tests revealed no significant differences in error between any of these groups using a significance threshold of 0.05 in all cases.

In this work, we aimed to enhance the precision of placement of the remodeled bone fragments during open cranial vault remodeling surgeries. We focused on the frontal bone and the SO bar to correct metopic craniosynostosis. To achieve this, we developed and compared three guidance methods: AR on a tablet, AR on Microsoft HoloLens 2, and 3D-printed spacers (only for the SO bar). We evaluated the three methods in two stages: first, using a 3D-printed patient-specific phantom in a simulated scenario, and then during the actual surgery. In all cases, we quantitatively analyzed the position of the bone fragments in relation to the VSP as a means to assess the effectiveness of each guidance method for this clinical application.

Several approaches have been proposed to objectify bone fragment positioning during craniosynostosis surgery before ours. For instance, Hochfeld et al. proposed using a stereotactic frame to control fragment position [29]. This device was not only invasive and complex to fit but also took around one hour to be set up, highly increasing surgical time. On the other hand, Kobets et al. described the use of intraoperative CT imaging to confirm the surgical outcome [30]. However, the acquisition of the CT scan not only increases the surgical time but also exposes the infant to an extra dose of ionizing radiation. In contrast, our proposal addresses many of these limitations. For instance, our AR solutions offer both an easy setup and real-time guidance, enabling surgeons to make multiple adjustments during the final positioning. Additionally, we used structured light scans to record surgical outcomes and evaluate the results, providing information comparable to CT scans without exposing the patient to radiation [31].

The three methods developed in this work provided similar levels of accuracy in bone placement, with no statistically significant differences between the data recorded during the simulation and the actual surgery. Based on these findings, the results obtained from the simulations are as valid for system validation as those from actual surgeries. This is crucial for clinical translation, as it suggests that our developed tools are valuable for their use in a real-world surgical environment, not just in a controlled simulation. Our data also revealed no statistically significant differences in placement accuracy when comparing the guidance methods, which means that none of them demonstrated clear superiority for accurately positioning the remodeled bone fragments.

In all cases, the mean translation errors obtained (Table 1) are low and comparable to those reported in our previous work, which were 0.70 ± 0.24 mm for the SO bar and 0.67 ± 0.33 mm for the frontal bone [22]. That earlier study was proof of concept for using AR on a tablet in this context and was validated using 3D-printed phantoms in a simulation scenario. The average rotation errors achieved with the three guidance methods presented in this work are also comparable to those obtained previously: 0.43 ± 0.30° for the SO bar and 0.39 ± 0.33° for the frontal bone. In our current work, all mean translation and rotation errors are below 1 mm and 1° in all scenarios and with all guidance methods, except for the rotation error achieved with the 3D-printed spacers in the surgical scenario when positioning the SO bar, where the error was 1.1 ± 0.7°.

In a deeper analysis of each method, 3D-printed spacers were the quickest and most intuitive approach, as they only needed to be inserted like puzzle pieces. Moreover, they have proven to be beneficial when digital technologies are not available or practical. However, their main limitation is that they are only suitable for guiding the positioning of the SO bar. We could not find an equivalent spacer for the frontal bone that would provide similar guidance without contacting the delicate brain. Moreover, although the translation and rotation results across all axes were not significantly different from those achieved with AR-based guidance methods, the 3D-printed spacers primarily guide the anteroposterior axis of the SO bar. Adjustments in the remaining axes must be made by the surgeon’s judgment. In contrast, both AR solutions offer more comprehensive information, displaying the exact 3D position the bone fragments should occupy. Although they require additional time for setup and the use of extra hardware to run the applications, the AR solutions might be a more reliable option in complex cases with potentially challenging corrections. Statistically significant differences were found for certain translation and rotation axes, but the error values have proven sufficiently low for this clinical application. If preferred, 3D-printed spacers and AR guidance could be combined, the former for quick initialization and the latter for finer tuning of the positioning. Still, if higher accuracy was desired, a third AR marker could be added to the setup to increase the working volume of the AR applications and, consequently, the perspectives from which the bone fragments can be verified.

Regarding AR in tablet versus Microsoft HoloLens, both devices were seamlessly integrated into the surgical room. Both devices effectively recognized the AR markers thanks to the consistent illumination of the surgical site. Moreover, their location was close enough to the region of interest to minimize registration error [32]. Notably, the placement of AR guides and markers can be easily adjusted for future patients to suit different surgical approaches. In our case, using the tablet was slightly more inconvenient than the Microsoft HoloLens 2 because it had to be handheld. However, the shared view it offered was valuable for collaborative decision-making in the OR. Additionally, the tablet’s display of virtual models was clearer and easier to interpret than that of the AR glasses, particularly for users less familiar with the latter visualization mode. Ultimately, the ease of use for each device depended largely on the user’s personal proficiency with the technology.

Notably, two surgeons used the Microsoft HoloLens 2 simultaneously during the surgery, each aligning the same bone from their own perspective. Although the applications were not synchronized, this approach was highly effective in maintaining the stability of the bone fragment from both sides. This method is advantageous over having a single surgeon using the AR device, which would require them to move around the patient multiple times, potentially becoming cumbersome and unstable.

This work develops and tests the feasibility and effectiveness of AR-based and 3D-printing-based solutions to enhance the precision of bone fragment placement during craniosynostosis surgeries. We designed three alternative guidance methods: AR on a tablet, AR with Microsoft HoloLens 2, and 3D-printed spacers, and initially evaluated them in a simulation scenario. Later, they were easily introduced in the surgical room and successfully employed to guide an actual surgery. The three guidance methods showed comparable accuracy in positioning remodeled bone fragments with mean placement errors below 1 mm in all cases.

While 3D-printed spacers are a practical solution when digital technologies are unavailable, AR-based methods provide more comprehensive guidance, especially in complex cases requiring meticulous adjustments. Nevertheless, our study suggests that the choice of method may depend more on the specific clinical context, user-specific skills, and available resources than on a clear superiority of one method over the others.

Looking forward, future research could focus on incorporating additional 2D AR markers or even 3D AR markers to increase the working volume of AR systems and obtain even greater accuracy. Additionally, we recommend the coupled use of the selected alternative so that surgeons can verify each other's work in real-time, thus improving the overall safety and efficacy of these advanced surgical techniques. Overall, the integration of AR and 3D printing into cranial surgery holds great promise for improving surgical precision and patient outcomes.

2D / 3D: Two / three-dimensional

AR: Augmented reality

EACS: Endoscopically assisted craniosynostosis surgery

FDM: Fused deposition modeling

HGUGM: Hospital General Universitario Gregorio Marañón

OCVR: Open cranial vault reconstruction

OR: Operating room

PLA: Polylactic acid

SDK: Software development kit

SLA: Stereolithography

SO bar: Supraorbital bar

VSP: Virtual surgical plan

Acknowledgements

The authors would like to thank David García Mato for his valuable contribution to the analysis of the optimal cranial shape model during the development of the virtual surgical plan for this patient.

Funding

This research was supported by projects PI22/00601 and AC20/00102 PerPlanRT (Ministerio de Ciencia, Innovación y Universidades, Instituto de Salud Carlos III, Asociación Española Contra el Cáncer and European Regional Development Fund “Una manera de hacer Europa”, ERA PerMed) and projects TED2021-129392B-I00 and TED2021-132200B-I00 (MCIN/AEI/10.13039/501100011033 and European Union “NextGenerationEU”/PRTR).

Ethics approval and consent to participate

The study received approval from the Research Ethics Committee at HGUGM and adhered to the principles of the 1964 Declaration of Helsinki, as revised in 2013.

Consent for publication

Written informed consent was obtained from the patient's parent for using their child's data for scientific purposes, including scientific publications.

Author’s contributions and competing interests

Conception and design of the work: A. P.-D.-d.-l.-L., M. G.-S., A. T., M. T., J.-V. D.-A., M. G. L., J. P., S. O.; Acquisition, analysis, and interpretation of data: A. P.-D.-d.-l.-L., M. G.-S., A. T., M. G. L., J. P., S. O; Software: A. P.-D.-d.-l.-L., M. G.-S.; Writing - original draft preparation: A. P.-D.-d.-l.-L., S. O.; Writing – review and editing: M. G.-S., J. P. All authors have read and agreed to the published version of the manuscript. They all have also agreed to be personally accountable for their contributions and to ensure that questions related to the accuracy or integrity of any part of the work are appropriately investigated, resolved, and the resolution documented in the literature.

Authors' information

Alicia Pose-Díez-de-la-Lastra: A. P.-D.-d.-l.-L.
Mónica García-Sevilla: M. G.-S.
Austin Tapp: A. T.
Manuel Tousidonis: M. T.
Juan-Vicente Darriba-Alles: J.-V. D.-A.
Marius George Linguraru: M. G. L.
Javier Pascau: J. P.
Santiago Ochandiano: S. O.

K. A. Klement, K. A. Adamson, N. L. Horriat, and A. D. Denny, “Surgical Treatment of Nonsyndromic Craniosynostosis,” Journal of Craniofacial Surgery, vol. 28, no. 7, p. 1752, Oct. 2017, doi: 10.1097/SCS.0000000000003950.
F. D. Rocco, E. Arnaud, and D. Renier, “Evolution in the frequency of nonsyndromic craniosynostosis,” Jul. 2009, doi: 10.3171/2009.3.PEDS08355.
H. Kabbani and T. S. Raghuveer, “Craniosynostosis,” afp, vol. 69, no. 12, pp. 2863–2870, Jun. 2004.
H. K. Delye, W. Borstlap, and E. Van Lindert, “Endoscopy-assisted craniosynostosis surgery followed by helmet therapy,” Surg Neurol Int, vol. 9, no. 1, p. 59, 2018, doi: 10.4103/sni.sni_17_18.
B. S. Lee et al., “Management options of non-syndromic sagittal craniosynostosis,” Journal of Clinical Neuroscience, vol. 39, pp. 28–34, May 2017, doi: 10.1016/j.jocn.2017.02.042.
D. R. Thompson, D. Zurakowski, C. M. Haberkern, P. A. Stricker, and P. M. Meier, “Endoscopic Versus Open Repair for Craniosynostosis in Infants Using Propensity Score Matching to Compare Outcomes: A Multicenter Study from the Pediatric Craniofacial Collaborative Group,” Anesthesia & Analgesia, vol. 126, no. 3, p. 968, Mar. 2018, doi: 10.1213/ANE.0000000000002454.
S. Arts, H. Delye, and E. J. van Lindert, “Intraoperative and postoperative complications in the surgical treatment of craniosynostosis: minimally invasive versus open surgical procedures,” Nov. 2017, doi: 10.3171/2017.7.PEDS17155.
R. H. Han et al., “Characterization of complications associated with open and endoscopic craniosynostosis surgery at a single institution,” Mar. 2016, doi: 10.3171/2015.7.PEDS15187.
L. Domínguez, C. Rivas-Palacios, M. M. Barbosa, M. A. Escobar, E. P. Florez, and E. García-Ballestas, “Outcomes of endoscopic treatment for early correction of craniosynostosis in children: a 26-year single-center experience,” Jun. 2023, doi: 10.3171/2023.4.PEDS22512.
J. A. Fearon, “Discussion: Nonsyndromic Unilateral Coronal Synostosis: A Comparison of Fronto-Orbital Advancement and Endoscopic Suturectomy,” Plastic and Reconstructive Surgery, vol. 143, no. 3, p. 849, Mar. 2019, doi: 10.1097/PRS.0000000000005384.
A. R. Porras et al., “Locally Affine Diffeomorphic Surface Registration and Its Application to Surgical Planning of Fronto-Orbital Advancement,” IEEE Transactions on Medical Imaging, vol. 37, no. 7, pp. 1690–1700, Jul. 2018, doi: 10.1109/TMI.2018.2816402.
A. R. Porras, D. Zukic, A. Equobahrie, G. F. Rogers, and M. G. Linguraru, “Personalized Optimal Planning for the Surgical Correction of Metopic Craniosynostosis,” in Clinical Image-Based Procedures. Translational Research in Medical Imaging, R. Shekhar, S. Wesarg, M. Á. González Ballester, K. Drechsler, Y. Sato, M. Erdt, M. G. Linguraru, and C. Oyarzun Laura, Eds., Cham: Springer International Publishing, 2016, pp. 60–67. doi: 10.1007/978-3-319-46472-5_8.
N. R. Saber et al., “Generation of normative pediatric skull models for use in cranial vault remodeling procedures,” Childs Nerv Syst, vol. 28, no. 3, pp. 405–410, Mar. 2012, doi: 10.1007/s00381-011-1630-7.
D. García-Mato et al., “Effectiveness of Automatic Planning of Fronto-orbital Advancement for the Surgical Correction of Metopic Craniosynostosis,” Plast Reconstr Surg Glob Open, vol. 9, no. 11, p. e3937, Nov. 2021, doi: 10.1097/GOX.0000000000003937.
D. M. Steinbacher, “Three-Dimensional Analysis and Surgical Planning in Craniomaxillofacial Surgery,” Journal of Oral and Maxillofacial Surgery, vol. 73, no. 12, pp. S40–S56, Dec. 2015, doi: 10.1016/j.joms.2015.04.038.
S. Soldozy et al., “Three-dimensional printing and craniosynostosis surgery,” Childs Nerv Syst, vol. 37, no. 8, pp. 2487–2495, Aug. 2021, doi: 10.1007/s00381-021-05133-8.
P. Udhay, K. Bhattacharjee, P. Ananthnarayanan, and G. Sundar, “Computer-assisted navigation in orbitofacial surgery,” Indian J Ophthalmol, vol. 67, no. 7, pp. 995–1003, Jul. 2019, doi: 10.4103/ijo.IJO_807_18.
M. Bruneau, F. Schoovaerts, R. Kamouni, S. Dache, O. De Witte, and S. de Fontaine, “The mirroring technique: a navigation-based method for reconstructing a symmetrical orbit and cranial vault,” Neurosurgery, vol. 73, no. 1 Suppl Operative, pp. ons24-28; discussion ons28-29, Sep. 2013, doi: 10.1227/NEU.0b013e318282a4e3.
D. García-Mato et al., “Craniosynostosis surgery: workflow based on virtual surgical planning, intraoperative navigation and 3D printed patient-specific guides and templates,” Scientific Reports, vol. 9, no. 1, pp. 1–10, 2019, doi: 10.1038/s41598-019-54148-4.
G. Coelho et al., “Augmented reality and physical hybrid model simulation for preoperative planning of metopic craniosynostosis surgery,” Mar. 2020, doi: 10.3171/2019.12.FOCUS19854.
A. Thabit, M. Benmahdjoub, M.-L. C. van Veelen, W. J. Niessen, E. B. Wolvius, and T. van Walsum, “Augmented reality navigation for minimally invasive craniosynostosis surgery: a phantom study,” Int J CARS, vol. 17, no. 8, pp. 1453–1460, Aug. 2022, doi: 10.1007/s11548-022-02634-y.
D. García-Mato et al., “Augmented reality visualization for craniosynostosis surgery,” Computer Methods in Biomechanics and Biomedical Engineering: Imaging and Visualization, vol. 9, no. 4, pp. 392–399, Oct. 2020, doi: 10.1080/21681163.2020.1834876.
W. Han et al., “A new method for cranial vault reconstruction: Augmented reality in synostotic plagiocephaly surgery,” Journal of Cranio-Maxillofacial Surgery, vol. 47, no. 8, pp. 1280–1284, Aug. 2019, doi: 10.1016/j.jcms.2019.04.008.
F. M. Anderson, “Treatment of Coronal and Metopic Synostosis: 107 Cases,” Neurosurgery, vol. 8, no. 2, p. 143, Feb. 1981.
S. Pieper, M. Halle, and R. Kikinis, “3D Slicer; 3D Slicer,” 2004 2nd IEEE International Symposium on Biomedical Imaging: Nano to Macro (IEEE Cat No. 04EX821), 2004, doi: 10.1109/ISBI.2004.1398617.
A. Díez-Montiel, A. Pose-Díez-de-la-Lastra, A. González-Álvarez, J. I. Salmerón, J. Pascau, and S. Ochandiano, “Tablet-based Augmented reality and 3D printed templates in fully guided Microtia Reconstruction: a clinical workflow,” 3D Printing in Medicine, vol. 10, no. 1, p. 17, May 2024, doi: 10.1186/s41205-024-00213-2.
J. A. Calvo-Haro et al., “Point-of-care manufacturing: a single university hospital’s initial experience,” 3D Printing in Medicine, vol. 7, no. 1, p. 11, Apr. 2021, doi: 10.1186/s41205-021-00101-z.
United States Pharmacopeial Convention, The United States Pharmacopeia. Rockville, MD, USA, 2012; Volume 1. Available online: https://www.usp.org/ (accessed on Feb 10, 2022), pp. 98, 1. [Online]. Available: https://www.usp.org/
M. Hochfeld, H. Lamecker, U.-W. Thomale, M. Schulz, S. Zachow, and H. Haberl, “Frame-based cranial reconstruction,” Mar. 2014, doi: 10.3171/2013.11.PEDS1369.
A. J. Kobets et al., “Virtual modeling, stereolithography, and intraoperative CT guidance for the optimization of sagittal synostosis reconstruction: a technical note,” Childs Nerv Syst, vol. 34, no. 5, pp. 965–970, May 2018, doi: 10.1007/s00381-018-3746-5.
D. García-Mato et al., “Intraoperative Outcome Evaluation in Craniosynostosis Reconstruction Surgery using 3D Photography,” in Proceedings of the 34th International Congress and Exhibition of Computer Assisted Radiology and Surgery, Munich: Springer Link, Jun. 2020, pp. S87–S88. doi: 10.1007/s11548-020-02171-6.
J. M. Fitzpatrick, J. B. West, and C. R. Maurer, “Predicting Error in Rigid-body, Point-based Registration,” IEEE Trans. Med. Imaging, vol. 17, no. 5, pp. 694–702, 1998, doi: 10.1109/42.736021.

No competing interests reported.

Download PDF

Journal Publication

published 21 Mar, 2025

Read the published version in 3D Printing in Medicine →

Editorial decision: Revision requested
14 Nov, 2024
Reviewers agreed at journal
15 Oct, 2024
Reviews received at journal
09 Oct, 2024
Reviews received at journal
09 Oct, 2024
Reviewers agreed at journal
25 Sep, 2024
Reviewers agreed at journal
25 Sep, 2024
Reviewers invited by journal
25 Sep, 2024
Editor assigned by journal
23 Sep, 2024
Submission checks completed at journal
23 Sep, 2024
First submitted to journal
16 Sep, 2024

You are reading this latest preprint version

Microsoft HoloLens 2 vs. Tablet-based Augmented Reality and 3D printing for fronto- orbital reconstruction of craniosynostosis: A case study

Status:

Journal Publication

Version 1

Abstract

Figures

1. Introduction

2. Materials and Methods

2.1. Clinical data

2.2. Virtual surgical planning and surgical guide design

2.3. Surgical workflow

2.4. Augmented reality reference markers and applications

2.5. 3D printing and sterilization

2.6. Performance evaluation

3. Results

4. Discussion

5. Conclusions

Abbreviations

Declarations

References

Additional Declarations

Supplementary Files

Status:

Journal Publication

Version 1