“A new breakthrough has been made in the field of medical image anomaly detection, and relevant experts have constructed a cross domain self supervised representation learning framework, effectively solving the problem of semantic differences in pre trained models in medical images, significantly improving the accuracy and robustness of anomaly detection, and providing a reliable solution for the development of this field.”