Adversarial learning-based domain adaptation algorithm for intracranial artery stenosis detection on multi-source datasets

Yuan Gao, Chenbin Ma, Lishuang Guo, Guiyou Liu, Xuxiang Zhang, Xunming Ji

Abstract: Highlights•ALDA is a deep learning method for automatic ICAS detection from RFIs.•Multi-source datasets enhance generalization of ICAS detection using RFIs.•ALDA leverages adversarial learning for feature representation sharing and distinguishability learning.•Integration of diverse datasets enhances generalizability and robustness of feature representations.•Efficient and generalizable method for early screening and diagnosis of ICAS using RFIs.