Fabrication of serum-based SERS-tailored 3D structures for thyroid cancer diagnosis.

Early detection of thyroid cancer improves patient survival rate from 51.9% to 99.9%. Fine needle aspiration cytology is the primary method for diagnosing thyroid cancer; however, this method is associated with limitations, including diagnostic uncertainty and potential complications. Despite numerous studies to identify diagnostic biomarkers for thyroid cancer, none has been found to date. Therefore, new methods that do not rely on biomarkers are warranted to aid thyroid cancer diagnosis. Here, we suggest a novel approach using 3D gold nanoclusters to obtain Surface-enhanced Raman scattering (SERS) spectra using the serum samples of patients with thyroid cancer and normal individuals. Briefly, an evaporation-based 3D printing technique was employed to fabricate nanoclusters containing serum. SERS spectra were collected from 50 normal individuals and 50 patients with thyroid cancer. The spectra were then analysed using machine learning with 1D and 2D convolutional neural networks (CNNs) architecture. Notably, the 2D CNN exhibited superior performance for the classification of thyroid cancer cases, with sensitivity of 93.1% and specificity of 84.0%. Such findings suggest the potential use of metabolite analysis for the diagnosis of thyroid cancer without finding biomarkers. This SERS measurement approach using 3D nanoclusters may also be leveraged for the diagnosis of other diseases.