Reconstructive surgery is commonly performed after ablative oncological resection in patients with head and neck cancer to restore function and improve quality of life. The fibula free flap (FFF) and the anterolateral thigh (ALT) flap are among the most commonly used free flaps for reconstruction in the head and neck region. Flap viability depends on an adequate vascular supply, which is provided by so-called perforators. Accurate identification of their location is crucial for determining the optimal position of the skin paddle of the flap.

Several techniques can be used for perforator localization, including the handheld Doppler, computed tomography angiography (CTA), and magnetic resonance angiography (MRA). However, these techniques have important limitations. Doppler does not provide information on the vessel’s size, course, length, origin, or its relationship with surrounding tissues. In contrast, CTA and MRA data can be transformed into three-dimensional (3D) models using image segmentation and 3D modeling techniques. These models allow visualization of the vessel’s size, course, length, origin, and its spatial relationship with surrounding tissues. However, a key limitation of CTA- and MRA-based 3D models is that they are not directly correlated with the patient’s physical anatomy during surgical planning of the flap.

Augmented reality (AR) may overcome this limitation by enabling spatial alignment of preoperative imaging data with the patient’s anatomy, thereby allowing more intuitive interpretation. This thesis introduces and evaluates an AR workflow for perforator visualization and localization in the FFF and ALT flap.

A workflow for AR projection of perforator 3D models onto the FFF and ALT donor sites is developed based on the results of experiments performed on a healthy volunteer. Subsequently, the newly developed workflow is validated in terms of registration accuracy in a second healthy volunteer. Finally, the workflow is validated in a patient case series in terms of registration accuracy and perforator localization accuracy, using intraoperative findings as the ground truth. The results showed that the target registration error (TRE) remained well below 10 mm for both flaps, a threshold that is generally considered clinically acceptable. However, with respect to intraoperative perforator localization accuracy, the FFF exceeded the 10 mm threshold, whereas the ALT flap remained well below this threshold. As these intraoperative findings are based on a small sample size, future studies should include a larger number of patients to enable more robust conclusions.

In addition, the clinical value and usability of the workflow are assessed. In this study, surgeons created surgical plans based on a conventional Doppler and two-dimensional (2D) MRA workflow, as well as an AR-guided workflow. The plans were compared, and usability was assessed using validated questionnaires. The results demonstrate that the AR-guided workflow enabled the design of smaller skin paddles and shorter incisions, thereby reducing donor site morbidity and potentially increasing the likelihood of primary closure. In addition, interobserver variability tended to decrease when the AR-guided workflow was used. These findings were accompanied by high perceived usefulness in favor of the AR-guided workflow, while the usability of both workflows was rated as moderate to high. Although these findings are promising, they are based on a limited dataset and a small number of participating surgeons. Furthermore, the study was conducted in an experimental rather than a clinical setting. Therefore, future studies involving larger cohorts and clinical studies are required to further validate these results.
... Read more