For patients that require hemodialysis, it is necessary to first establish a dialysis fistula that connects the patient's artery and vein, such as in the patient's arm. The location of this dialysis fistula and nearby vessel structure has to be communicated by the hospital department that installs the dialysis fistula to the department that performs hemodialysis, which tend to be imprecise because the vessels and the fistula are under the skin and not visible from outside. We propose a technique that use 3D computer vision techniques to allow the augmented-reality visualization of the vessels from outside by computing and aligning the 3D models of the vessels and the patient's arm surface. In addition to demonstrate this technique, we also perform evaluation of the vessel localization accuracy using a wire as a simulated vessel, and find the accuracy to be sufficient for the technique to be useful in practice.