Improving Terahertz Hyperspectral Image Analysis in the Frequency- and Time-Domain

Terahertz time-domain spectroscopy (THz-TDS) is a strong candidate for biological imaging applications due to the non-destructive, non-ionizing nature of THz radiation, coupled with the THz-range spectral fingerprints of many molecules. One of the primary challenges within THz imaging is effective analysis of the large datasets generated by THz imaging. We present an automated method to differentiate unique sample components through THz hyperspectral image analysis. Standard hyperspectral image analysis for THz-TDS data relies on extracting the amplitude of the frequency-domain transmission spectra. We demonstrate improved ability to differentiate sample features by incorporating both amplitude and phase components of THz transmission data into analysis methods. Furthermore, we show the ability to achieve comparable sample component separation through analysis performed on the native time-domain THz data. Eliminating the Fourier transformation from the data analysis process reduces computational cost and time while maintaining integrity of the results with respect to sample composition. To illustrate both the frequency- and time-domain analysis methods, we show the automatic separation of structural components within a split-ring resonator metamaterial sample.