Distinguishing different proteins based on terahertz spectra by visual geometry group 16 neural network

Detecting different kinds of proteins is of great significance for medical diagnosis, biological research, and other fields. We combine both terahertz (THz) absorption and refractive index spectra with the visual geometry group 16 (VGG-16) neural network to intelligently identify four proteins, namely albumin, collagen, pepsin, and pancreatin in this study. The THz absorption-refractive index spectra of the proteins were converted to two-dimensional image features by the Grassia angular summation field (GASF) method and used as a dataset, which enabled the VGG-16 model to achieve 98.8% accuracy in distinguishing the four proteins. We also compared the VGG-16 model with other machine learning models, which demonstrate that it has better performance. Overall, the VGG-16 neural network transfer learning technique proposed in this study can quickly and accurately achieve the identification of different kinds of proteins. This research might have potentially important applications in biotechnology fields, such as biosensors, biopharmaceuticals, and medicine.