Enhancing Segmentation Accuracy of the Common Iliac Vein in OLIF51 Surgery in Intraoperative Endoscopic Video through Gamma Correction: A Deep Learning Approach.

While minimally invasive surgery reduces the physical burden on the patient, the lack of depth perception and the difficulty in grasping the position and morphology of organs make it a burdensome procedure for the surgeon. In particular, when an endoscope is used to secure the field of view, two-dimensional images are used, making it difficult to fully grasp the complex three-dimensional structures and identify the organs. This problem is no exception in the treatment of spinal disorders. In lumbosacral anterolateral intervertebral fusion, which can treat a variety of spinal disorders, the common iliac vein (CIV), a type of vena cava, runs in the vicinity of the operative field. This vein has a fragile, thin wall and carries large volumes of blood at low pressure, which can cause fatal hemorrhage if injured intraoperatively. However, they are often located behind the peritoneum or arteries and are difficult to identify because of their similar redness to the surrounding tissues. This increases the risk of damage caused by surgical instruments during field development and intraoperative manipulation. In this technique, an assistive endoscope is used to secure the field of view. The surgeon compares the 3D construction images (MRI, CT, etc.) taken preoperatively with the actual space to ensure an appropriate surgical route. However, the identification of veins depends on the surgeon's experience, and only a limited number of physicians are able to perform this procedure safely and reliably. Therefore, an intraoperative support system is required to standardize the identification, but such a system has not been established. Therefore, we developed a method for automatic extraction of vein regions in endoscopic images to facilitate identification and reduce the risk of intraoperative injury by visually presenting the common iliac vein to the surgeon.