Highly Efficient Blood Protein Analysis Using Membrane Purification Technique and Super-Hydrophobic SERS Platform for Precise Screening and Staging of Nasopharyngeal Carcinoma

Abstract

Early screening and precise staging are crucial for reducing mortality in patients with nasopharyngeal carcinoma (NPC). This study aimed to assess the performance of blood protein surface-enhanced Raman scattering (SERS) spectroscopy, combined with deep learning, for the precise detection of NPC. A highly efficient protein SERS analysis, based on a membrane purification technique and super-hydrophobic platform, was developed and applied to blood samples from 1164 subjects, including 225 healthy volunteers, 120 stage I, 249 stage II, 291 stage III, and 279 stage IV NPC patients. The proteins were rapidly purified from only 10 µL of blood plasma using the membrane purification technique. Then, the super-hydrophobic platform was prepared to pre-concentrate tiny amounts of proteins by forming a uniform deposition to provide repeatable SERS spectra. A total of 1164 high-quality protein SERS spectra were rapidly collected using a self-developed macro-Raman system. A convolutional neural network-based deep-learning algorithm was used to classify the spectra. An accuracy of 100% was achieved for distinguishing between the healthy and NPC groups, and accuracies of 96%, 96%, 100%, and 100% were found for the differential classification among the four NPC stages. This study demonstrated the great promise of SERS- and deep-learning-based blood protein testing for rapid, non-invasive, and precise screening and staging of NPC.

Keywords: deep learning; nasopharyngeal carcinoma; protein SERS; super-hydrophobic platform.

Figure 1

(A) Schematic of the procedure for preparing the super-hydrophobic platform by chemical etching and chemical vapor deposition. (B) Photograph of Al sheet-based super-hydrophobic platform. (C) SEM micrograph of the super-hydrophobic grooved surface. (D) Optical image of a water droplet on the surface. (E) Photograph and TEM micrograph of the Ag NPs. (F) Ultraviolet–visible (UV) absorption spectrum of the Ag NPs. (G) Zeta potential of the Ag NPs.

Figure 2

(A) Schematic of CA-membrane-based purification of plasma protein for SERS detection. (B) High-throughput, rapid macro-Raman system (abbreviations: L, lens; SP, short pass; LP, long pass). (C) Protein SERS spectra for NPC screening and staging using a deep-learning algorithm, and (D) structure of CNN-based deep-learning model (abbreviations: Conv, convolutional layer; FC, fully connected layer; N, normal; I, stage I NPC; II, stage II NPC; III, stage III NPC; IV, stage IV NPC).

Figure 3

(A) Comparison of a SERS spectrum from plasma protein purified by CA membrane, a SERS signal from a blank CA membrane, and the background signal from the Ag NPs. (B,C) Spectra from five random sampling points along the diameters of the dried mixtures on the super-hydrophobic and the ordinary Al surfaces, respectively. (D) Comparison of the normalized average protein SERS spectra from healthy and all NPC subjects, and the corresponding difference spectrum (NPC minus normal). (E) Average protein SERS spectra from NPC subjects from the four stages, and (F) their corresponding difference spectra.

Figure 4

Box plots of the peak intensities of protein SERS spectra for the normal volunteers and NPC subjects at all stages. For each box, the central line represents the median, and the lower and upper boundaries indicate the 25th and 75th percentiles, respectively. Abbreviations: ns indicates no significance; ** p < 0.01; *** p < 0.001; **** p < 0.0001 (Mann–Whitney U test).

Figure 5

Box plots of some of the protein SERS peak intensities for NPC subjects at each of the four stages. Abbreviations: ns indicates no significance; * p < 0.05; ** p < 0.01; *** p < 0.001; **** p < 0.0001 (Kruskal–Wallis test).

Figure 6

(A) Loss and accuracy curves of the CNN model on the training set. (B) Training accuracy and (C) testing accuracy of the CNN method for healthy and stage I, II, III, and IV NPC groups on the training and test set, respectively. (D) Scatter plot of the discrimination scores by PCA-LDA on the training set. (E) Training accuracy and (F) testing accuracy of PCA-LDA for the five groups.