Prospective study on the localization of anterolateral thigh perforator vessel based on mixed reality and artificial algorithm.

Abstract

Objectives: This paper aims to construct a system integrating mixed reality technology with artificial algorithm and to evaluate its effectiveness in vascular localization during anterolateral thigh perforator flap surgery to provide new insights for clinical practice.

Methods: Twenty patients undergoing anterolateral thigh perforator flap repair were selected. After attaching positioning devices on the lower limb, CT angiography (CTA) scans were performed. The 2D data obtained were converted into a 3D model of the positioning device and vessels. Mixed reality technology was utilized to achieve 3D visualization of perforator vessels. An artificial algorithm was developed in HoloLens 2 to match the positioning device automatically with its 3D model intraoperatively to overlap the perforator vessels with their 3D models. The number of perforator vessels identified within the flap harvesting area and the actual number detected during surgery were recorded to calculate the accuracy rate of vessel identification based on CTA data reconstruction. The distance between the perforator vessel exit points located by the system and the actual exit points was measured, and the error values were calculated. The surgical time required for the system to harvest the anterolateral thigh perforator flap was documented and compared with the surgical time required by conventional methods. The clinical applicability of the system was discussed.

Results: The CTA data reconstruction identified 30 perforator vessels, while the actual number found during surgery was 32, resulting in an identification accuracy rate of 93.75%. The average distance between the perforator vessel exit points located by the system and the actual exit points was (1.65±0.52) mm. The average surgical time for flap harvesting with the assistance of the system was (43.45±4.6) min compared with (57.6±7.9) min required by conventional methods. All perforator flaps survived the procedure. One case of flap infection occurred seven days postoperatively, and one case of partial flap necrosis was treated with symptomatic therapy, resulting in delayed healing.

Conclusions: The system constructed in this paper can achieve 3D visualization of perforator vessels through mixed reality technology and improve the accuracy of perforator vessel localization using artificial algorithms, hence demonstrating potential application in anterolateral thigh perforator flap harvesting surgeries.