FOA is one of the most prevalent surgical interventions for addressing both syndromic and non-syndromic craniosynostosis. These procedures rely on surgeons' skills and judgment. Consequently, innovations are being developed to enhance the surgical workflow. One existing method involves creating personalized resorbable osteosynthesis plates to reduce intraoperative adjustments [8]. Clinically, resorbable plates have shown good outcomes and benefits in pediatric craniofacial surgery [24, 25] and their use inside the skull’s cortex is a safe and effective method, avoiding plate contour prominence [25].

Another approach that optimizes the osteotomy procedure is through VSP and 3D-printed patient-specific surgical cutting guides, producing reproducible results for FOA surgery [15]. Nonetheless, next to the time-consuming process of surgical template design, a perfect fit of the guide could be difficult to ensure. Problems in geometric fit could arise from guide design errors, manufacturing inaccuracies, or soft tissue obstruction. Therefore, robot-guided laser osteotomy has emerged as another precise method. The CARLO® system’s performance has been investigated in multiple studies, showing its potential as a precise tool to replace conventional instruments for bone surgery [19, 20, 26, 27]. This study assessed FOA accuracy using resorbable PSIs and robot-guided laser osteotomy for FOA in a complete digital workflow.

Implant production at the point-of-care allows flexibility for last-minute changes. This study was conducted at the University Hospital Basel's 3D Print Lab using the Arburg Freeformer 200-3X, a printer for research on biodegradable implant fabrication [28]. Printing six implants for one skull took about 3.5 h. Post-processing involved dissolving support material and drilling screw holes. Surface scan analysis of the 3D-printed PSIs showed minor deviations from the intended designs, with surface deviations ranging from 0.06 to 0.20 mm (Table 2). The largest deviation was in the frontal plate, potentially due to thermal expansion or warping during printing. Using APF technology and pre-drilled screw holes by CARLO®, the 3D-printed implants facilitate precise bony fragment advancement in surgery, eliminating the need for custom positioning templates. Another advantage is the ability to create highly customized implant designs without manual adaptation. However, the biocompatibility, resorption, and mechanical behavior of sterilized PLDLLA/TCP for FOA need assessment, along with compatibility with common bone fixation systems like SonicWeld by KLS Martin.

Our study demonstrated that the robot-guided laser osteotomy grooves had a median deviation from the preoperative paths of approximately 0.44 mm, which is in a similar range as previously reported by Holzinger et al. [15] with the same system. In our study, a maximum deviation of 1.7 mm was observed, possibly due to registration errors from the camera navigation system or deformation of the plastic skulls under stress of the position marker clamp. Shape accuracy was assessed through surface scans (Fig. 6a, Table 2), showing a mean deviation of 0.04 mm for the 3D-printed skulls, with selective laser sintering possibly contributing to this error. The osteotomy deviation of the CARLO® is comparable to the linear AR-guided procedure described by García-Mato et al., who reported medial errors of < 1 mm for craniosynostosis osteotomies. Just like robot-guided procedures, AR seems to be able to provide accurate intraoperative guidance, providing real-time visual overlays of the VSP. AR systems are typically more accessible and less bulky than robot-guided systems; however, they rely on the surgeon’s manual precision to perform the osteotomy, introducing variability. Robot-guided systems offer automated execution, making them ideal for complex craniosynostosis procedures which require consistent accuracy. The integrated OCT system of the robot-guided osteotome enables precise depth control [26, 29]. However, considering the risk of neurological damage due to the proximity of meninges and brain tissue is crucial. The safety and accuracy of the depth measurement system of CARLO® primo +  requires further evidence through cadaver studies and pre-clinical trials.

One of the key challenges with navigation systems for pediatric cranial surgery is the difficulty to achieve rigid head fixation in infants with unfused sutures, open fontanelles and thin cranial bones, which also complicates the attachment of tracking reference tools for the navigation system. A potential solution for marker placement is the Modified Mayfield Rubber Stopper Technique, which employs soft rubber stoppers over Mayfield pins to evenly distribute pressure, thereby minimizing the risk of cranial injury [30] while providing a solid frame on which the optical markers are attached. This technique offers a less invasive approach to head stabilization, ensuring both safety and accurate tracking for navigation for pediatric cranial procedures.

The robot-guided laser was occasionally limited by the orbital cavity, necessitating manual cuts around this area which could have affected the overall accuracy of FOA. In the operating room, the robot’s large head and footprint, expenses, and the necessity to preserve delicate tissues prevent it from completely replacing conventional piezoelectric devices or drilling tools. In routine FOA procedures, incomplete osteotomies of the supraorbital bar are performed to achieve the desired cranial shape by bending the bone [31]. In these cases, the bone segments remain connected in non-expanded locations, allowing bending to the ideal position using the plate as a guide, improving stability between the bone segments. Likewise, instead of performing complete osteotomies, we propose that the pulse layers of the laser osteotomy could in future be reduced to merely indicate the cutting planes, enabling surgeons to complete osteotomies with conventional cutting devices, thereby reducing operational time. The osteotomy procedure lasted approx. three hours. Time is crucial in pediatric surgeries as longer procedures increase the risk of complications like blood loss and anesthesia-related issues. With improvements in navigation- and robotic systems and a better understanding of the needs of the patient and medical practitioners, we expect the operational time of the osteotomies to decrease, which will streamline the process and significantly reduce operative time, enhancing efficiency in future surgeries.

Despite various procedural complexities, the final median surface deviation of the FOA was relatively low, remaining within the range of 1.01–1.35 mm (Table 3). The maximum deviation of 3.51 mm around the orbital region likely resulted from the limited stability of the frontal plate, contributing to the medial rotation of bony segments lateral to the orbits. An in vivo median accuracy within this range is considered clinically acceptable, although, to our knowledge, no previous studies have quantitatively compared the accuracy of conventionally performed FOAs. Future research should investigate the impact plate design to optimally support repositioned bone segments.

Using the proposed workflow, the need for surgical template planning can be omitted, saving significant pre-operative planning time. Additionally, an efficient implant design reduces the number of screws needed, decreasing costs. We believe this technology will enhance the efficiency of surgical interventions, improve surgical outcomes, and reduce the risk of complications and revision surgeries.