Presentation Information
[8p-PA2-20]Terahertz Spectroscopy for Accurate Polymorph Identification Enhanced by Machine Learning
〇(M1)Kazuyuki Morone1, Wangxuan Zhao1, Verdad C. Agulto1, Kosaku Kato1, Mihoko Maruyama2, Makoto Asakawa3 (1.ILE, Univ. Osaka, 2.Grad. Sch. of Eng., Univ. Osaka, 3.Kansai Univ.)
Keywords:
terahertz spectroscopy,machine learning,urinary stone
An automated identification method for urinary stone components using terahertz time-domain spectroscopy (THz-TDS) combined with machine learning was investigated. Classification and concentration estimation were performed using the absorption coefficient spectra of calcium oxalate monohydrate (COM) and calcium oxalate dihydrate (COD), demonstrating highly accurate identification. Furthermore, the performance of deep learning and Random Forest models was compared under limited-data conditions, and the effectiveness of data augmentation was also evaluated.