Image Processing Based Autonomous Guidewire Navigation in Percutaneous Coronary Intervention
Yong‐Jun Cho, Jae-hyeon Park, Jaesoon Choi, Dong Eui Chang
Abstract
Percutaneous coronary intervention (PCI) is a frequently used surgical method for infamous cardiovascular diseases such as angina pectoris and myocardial infarction. In PCI, a doctor inserts a guidewire into a patient’s blood vessel, watches angiogram images of the blood vessel, and controls the guidewire to the target point in the blood vessel. Recently, many researches have been conducted to automatically navigate a guidewire without a doctor’s effort. However, many of the researches need additional sensors or modification to the guidewire. Also, there have been attempts to apply reinforcement learning in autonomous PCI guidewire navigation. However, these attempts need further validation of their algorithms in various blood vessel structures. In this research, we develop an automatic control method that navigates a PCI guidewire using only image processing algorithms with no additional sensors nor modification to the guidewire. We validate our PCI guidewire navigation method in blood vessel models of SOFA simulator and on a real test bench. We achieve a top success rate in our experiment. Our PCI guidewire navigation videos are available on the internet to help readers’ comprehension.