Automated precision localization of peripherally inserted central catheter tip through model-agnostic multi-stage networks

Subin Park, Yoon Ki Cha, Soyoung Park, Myung Jin Chung, Kyung-Su Kim

Abstract: Highlights•We proposed a deep learning-based framework to solve the multiple fragments phenomenon.•The multi-stage network can provide to resolving false-positives or false-negatives.•Our framework reduced the average error by more than 50% compared to existing deep learning.•Our study will prevent complications from abnormal detection of PICC tip position.