Label-Free SERS Analysis of Serum Using Ag NPs/Cellulose Nanocrystal/Graphene Oxide Nanocomposite Film Substrate in Screening Colon Cancer

• Published in: Nanomaterials (Basel, Switzerland) Switzerland), 2023-01, Vol.13 (2), p.334
• Main Authors: Li, Jie, She, Qiutian, Wang, Wenxi, Liu, Ru, You, Ruiyun, Wu, Yaling, Weng, Jingzheng, Liu, Yunzhen, Lu, Yudong
• Format: Article
• Language: English
• Subjects: Biocompatibility; Cancer screening; Cellulose; Cervical cancer; Colon cancer; Colorectal cancer; Diagnosis; Discriminant analysis; Free surfaces; Graphene; graphene oxide; label-free; Lasers; Machine learning; Medical prognosis; Microscopy; Nanocomposites; Nanocrystals; Nanoparticles; Patients; PCA-LDA; Principal components analysis; Raman spectra; Reproducibility; Screening; Self-assembly; Sensitivity; SERS; Silicon wafers; Silver; Spectrum analysis; Spin coating; Substrates
• ISSN: 2079-4991; 2079-4991
• DOI: 10.3390/nano13020334

Label-free surface-enhanced Raman scattering (SERS) analysis shows tremendous potential for the early diagnosis and screening of colon cancer, owing to the advantage of being noninvasive and sensitive. As a clinical diagnostic tool, however, the reproducibility of analytical methods is a priority. Herein, we successfully fabricated Ag NPs/cellulose nanocrystals/graphene oxide (Ag NPs/CNC/GO) nanocomposite film as a uniform SERS active substrate for label-free SERS analysis of clinical serum. The Ag NPs/CNC/GO suspensions by self-assembling GO into CNC solution through in-situ reduction method. Furthermore, we spin-coated the prepared suspensions on the bacterial cellulose membrane (BCM) to form Ag NPs/CNC/GO nanocomposite film. The nanofilm showed excellent sensitivity (LOD = 30 nM) and uniformity (RSD = 14.2%) for Nile Blue A detection. With a proof-of-concept demonstration for the label-free analysis of serum, the nanofilm combined with the principal component analysis-linear discriminant analysis (PCA-LDA) model can be effectively employed for colon cancer screening. The results showed that our model had an overall prediction accuracy of 84.1% for colon cancer ( = 28) and the normal ( = 28), and the specificity and sensitivity were 89.3% and 71.4%, respectively. This study indicated that label-free serum SERS analysis based on Ag NPs/CNC/GO nanocomposite film combined with machine learning holds promise for the early diagnosis of colon cancer.