Rapid quantitative detection of S100B based on nanophotonic heterochain chip and deep learning:methodological development,performance verification,and preclinical application guidelines
WANG Yuxin
LÜ Wenying
CHENG Gang
GAO Zhao
LI Yanteng
SUN Junzhao
WANG Peng
GUO Baorui
YANG Fan
ZHANG Rui
HAO Fangbin
SU Shichao
HE Renke
LIU Congwei
YIN Zhiyong
WANG Yumeng
LIU Jiayu
ZHANG Jianning
Abstract：Objective Based on existing research,and addressing the need for ultra-early detection of brain injury biomarkers,this study systematically summarizes and standardizes a rapid quantitative detection method for S100B utilizing nanophotonic heterochain chips combined with deep learning-based image recognition.It reports key analytical performance indicators and quality control procedures,and explores its preclinical application pathway.Methods Focusing on the five components of"chip-imaging-AI-quantification-quality control",the platform's detection process and parameter settings were outlined.This specifically included:standard sample evaluation of dynamic range and limit of detection,consistency comparison with traditional ELISA,assessment of interferents/cross-reactivity,and analysis of response time and field applicability.For real samples,methodological validation was based solely on detectability and consistency,without involving comprehensive clinical statistical results.Based on the above findings,a standardized operating procedure(SOP),a quality control checklist,and application pathway recommendations suitable for frontline/emergency/intraoperative scenarios were developed.Results During standard sample validation,the platform demonstrated a wide dynamic range and good fitting,with intra-batch/inter-batch variations controlled within acceptable levels,achieving stable readings within 15-30 min.It showed no significant cross-reactivity with common nerve-related proteins.For real samples,the platform exhibited high detectability for low-concentration samples that were undetected by traditional methods,and maintained good consistency with the reference method.Conclusion This methodological system possesses the comprehensive advantages of being applicable for ultra-early stage,low-abundance detection,and field deployment.The supporting SOP and key quality control points facilitate its standardized promotion and application in emergency and perioperative settings.Subsequent multicenter real-world studies are recommended to further accumulate translational evidence.
Keywords：S100Bnanophotonicsbiosensordeep learningtraumatic brain injurypoint-of-care testingartificial intelligenceimage analysis
Publication Date：2026-02-28
Online Publishing Date：2026-08-26(First online date of this platform, not the publication date of the document)
Pages：6( 218-223 )