Document is current

Article History

Received: 27 November 2024

Accepted: 31 March 2025

First Online: 15 May 2025

Declarations

:

: The authors declare no competing interests.

: This study utilized publicly available medical image datasets for training, testing, and validation. All medical data were fully anonymized, devoid of identifiable patient information such as names, addresses, or birthdates, and usage complied with institutional data policies, adhering to the intended scope and purpose. To ensure sufficient diversity and reduce potential biases or negative impacts, we used multiple public medical datasets. Extensive ablation and comparative experiments were conducted to evaluate the quality of these public datasets, ensuring data accuracy and authenticity.