Assessment of vessel deformation using deep learning-based semantic segmentation algorithm during needle manipulation in microvascular anastomosis: A pilot study

Abstract

Abstract Appropriate needle manipulation to avoid abrupt deformation of fragile vessels is a critical determinant of the success of microvascular anastomosis. However, no studies have evaluated the shape and area of the surgical objects. The present study aimed to develop a deep learning-based semantic segmentation algorithm and to assess vessel deformation in microvascular anastomosis for objective surgical skill assessment of "respect for tissue.” Semantic segmentation algorithm was trained based on a ResNet-50 network using microvascular end-to-side anastomosis training videos with artificial blood vessels. Using the created model, tissue deformation was analyzed, and the threshold violation error numbers were compared between expert and novice surgeons during the completion task of one stitch. High validation accuracy (99.1%) and Intersection over Union (0.93) were obtained for the auto-segmentation model. While completing the one-stitch task, experts showed significantly fewer errors than novices (p < 0.001), with a shorter completion time (p < 0.001). Significant differences were also observed in the phase of needle insertion (p = 0.04) and needle extraction (p < 0.001) between experts and novices. In conclusion, the assessment of vessel deformation during microvascular anastomosis using a deep-learning-based semantic segmentation algorithm is presented as a novel concept for evaluating microsurgical performance. This will be useful for future computer-aided devices to enhance surgical education and patient safety.