An intelligent serological SERS test toward early-stage hepatocellular carcinoma diagnosis through ultrasensitive nanobiosensing

Early or very early detection of hepatocellular carcinoma (HCC) is an effective means to resolve the low cure rates, but there currently lacks a method that fulfills clinical requirements. One of the most prospective approaches to detecting early-stage HCC is directly testing a compendium of disease-relevant biomolecules contained within human serum through surface-enhanced Raman scattering (SERS) nanobiosensing and recognizing the biomolecular patterns. We report a novel Si-based bimetallic nanoensembles-functionalized SERS substrate (its analytical enhancement factor reaches 1.47 × 1012) and introduce an ultrasensitive nanobiosensing for capturing the subtle characteristic changes in SERS spectra caused by HCC, hepatitis B, or cirrhosis. Toward early diagnosis, we created an intelligent serological test with this nanobiosensing and the deep learning algorithm to gain key biomolecular phenotypes of early-stage HCC. Using clinical samples from four target populations (normal, HCC, cirrhosis, and hepatitis B), the proof-of-principle result indicates that the test yielded a predictive accuracy of 98.75% on a held-out dataset (randomly drew 4 out of 28 samples per population). On the same held-out dataset, the sensitivity and specificity of the test were both 100% for distinguishing HCC. Such a new-concept liquid biopsy could provide an opportunity for early diagnosis of HCC.