Deep Learning-Based Spectroscopic Analysis
What would happen when a laser is irradiated on a certain material? Light will interact with the material so that we can detect some optical spectra such as IR spectrum or Raman spectrum. This optical signal can be called a molecular fingerprint; containing significant information derived from the molecular structure of the material. If we can detect the fingerprint information of biomarkers in our body, it will be possible to diagnose a disease or identify an unknown virus. However, since the biomarkers are compounds of complex molecules, it is challenging to interpret and analyze the molecular fingerprints. In order to overcome the problem, widely emerging statistical methods and machine learning techniques can be utilized. Our laboratory is now conducting research to diagnosis cancer and other diseases by developing optimized artificial intelligence for analyzing optical spectra from biomarkers.
[Representative papers]
-“Differential Diagnosis of Lung Cancer Using Artificial Intelligence-Based Extracellular Vesicle Analysis: A Pilot Study”, ACS Nano (2026) -“Needle-Free Prediction of Fetal Lung Maturity Using Vaginal Fluid Extracellular Vesicles ”, JAMA Pediatrics (2026) -“Diagnosis of Primary Open-Angle Glaucoma Using Spectral Profiling of Aqueous Humor-Derived Exosomes”, ACS Sensors (2026)
