A Quantitive Description Method of Vascular basing on Unsupervised Learning towards Operation Skills Assessment of Endovascular Surgery

Abstract

In the field of rapidly developing endovascular technique and technology, accurate assessment of surgical operation is essential for improving the efficiency of endovascular surgery and the performance of endovascular surgery robots. Existing methods of assessment have taken into consideration of a variety of indicators such as path length of operation, operation time and so on. The indicators that have been considered all come from surgeon's operation itself. However, the characteristics of specific patients' blood vessels are not considered for objective assessment. So in this paper, operating difficulty of different blood vessels was described for operation through the aortic arch by machine learning k-means models. Then clustering results were verified with external and internal metrics. Based on this study, difficulty levels of blood vessels can be taken as an important indicator for surgeons' endovascular operation evaluation in the future research.