| Popis: |
To distinguish clinically signicant fungus, Fourier transform infrared spectroscopy (FTIR) was used. In this work, 75 Trichosporon strains from ve different species were cultivated on SDA media and FTIR attenuated total reection (ATR) readings was taken. The classication (FTIR spectra) results of cluster analysis were compared to articial neural network (ANN) analysis (supervised approach). Validation of training set showed that both techniques properly categorized 100% of the spectra, at least for T. asahii (n = 62) and T. inkin (n = 8). With the addition of T. loubieri (n=1) and T. asteroids (n=1), the ANN's accuracy became reliant on the training database, resulting in 90% to 100% classication. |
