Non-invasive microwave detection system for rapid detection of intracerebral hemorrhage based on hydrogel metasurfaces and machine learning.

To develop a portable microwave-based detection device integrating hydrogel metasurface technology and machine learning for non-invasive rapid detection of intracerebral hemorrhage. By optimizing electromagnetic wave propagation paths through hydrogel metasurfaces to reduce cranial attenuation of electromagnetic signals and enhance transmission efficiency, combined with machine learning algorithms to analyze complex electromagnetic signatures, the system enables accurate diagnosis of intracerebral hemorrhage. In murine models, the system demonstrated 100% sensitivity, 84.21% specificity, and 93.75% accuracy and an AUC of 0.99 for detecting a 30-μL intracerebral hemorrhage. The incorporation of hydrogel metasurface improved the detection accuracy by 7.8%. This research provides a technical approach for early intracerebral hemorrhage diagnosis, demonstrating advantages including portability, rapid response and high sensitivity. The methodology promises to provide clinically effective early diagnostic solutions for cerebral hemorrhage patients.