USG-Net: Deep Learning-based Ultrasound Scanning-Guide for an Orthopedic Sonographer

초록

Ultrasound (US) imaging is widely used in the field of medicine. US images containing pathological information are essential for better diagnosis. However, it is challenging to obtain informative US images because of their anatomical complexity, which is significantly dependent on the expertise of the sonographer. Therefore, in this study, we propose a fully automatic scanning-guide algorithm that assists unskilled sonographers in acquiring informative US images by providing accurate directions of probe movement to search for target disease regions. The main contributions of this study are- (1) proposing a new scanning-guide task that searches for a rotator cuff tear (RCT) region using a deep learning-based algorithm, i.e., ultrasound scanning-guide network (USG-Net); (2) constructing a dataset to optimize the corresponding deep learning algorithm. Multidimensional US images collected from 80 patients with RCT were processed to optimize the scanning-guide algorithm which classified the existence of RCT. Furthermore, the algorithm provides accurate directions for the RCT, if it is not in the current frame. The experimental results demonstrate that the fully optimized scanning-guide algorithm offers accurate directions to localize a probe within target regions and helps to acquire informative US images.

이경수

조교수

연합학습 및 표현 학습을 사용한 medical 도메인에서의 딥 러닝 기법, 도메인 적응 및 테스트 타임 학습과 같은 딥 러닝 기반 컴퓨터 비전 응용, 이미지 처리 및 이미지-텍스트 캡셔닝을 포함한 의료 응용의 딥 러닝 기반 진단 등을 포함합니다.