研究目的
To set up a SERS-based database of DNA for AI-based sensing applications, enabling label-free discrimination of tumor suppressor genes.
研究成果
The silicon nanohybrids chip-based SERS method is effective for setting up SERS big data, allowing AI-based DNA discrimination in label-free manners with high accuracy. This approach could be useful for the production of massive and reliable database for AI sensing applications.
研究不足
The accuracy of DNA discrimination could be further improved with more representative SERS spectra for AI training and more optimized parameters in DNN.
1：Experimental Design and Method Selection:
The study employs a silicon nanohybrids chip made of silver nanoparticles (Ag NPs) in situ grown on silicon wafer surface (Ag NPs@Si) for SERS detection, integrated with a deep neural network (DNN) for data analysis.
2：Sample Selection and Data Sources:
Three kinds of representative tumor suppressor genes (p16, p21, and p53 fragments) with different base lengths are selected for SERS spectra collection.
3：List of Experimental Equipment and Materials:
Ag NPs@Si SERS chip, Raman microscope (HR800, Horiba Jobin Yvon, France), He-Ne laser (λex = 633 nm).
4：Experimental Procedures and Operational Workflow:
DNA samples are dropped onto the Ag NPs@Si substrates, dried, and measured with Raman mode. SERS spectra are collected and used as input data for DNN training.
5：Data Analysis Methods:
The DNN model is trained using back-propagation (BP) algorithm, with SERS data preprocessed to extract main eigenvalues for training.
独家科研数据包，助您复现前沿成果，加速创新突破
获取完整内容
