On the textile fibre’s analysis for forensics, utilizing FTIR spectroscopy and machine learning methods

Fibres are prevalent and can be encountered as trace evidence in various situations. In cases of rape and physical assault, analyzing trace fibre components and assessing their transferability can establish connections between individuals and crime scenes or between perpetrators and victims. This study involved Attenuated Total Reflectance – Fourier Transform Infrared (ATR–FTIR) characterization of 104 fibre samples, including natural fibres like cotton and wool (43 samples) and terry wool and synthetic fibres (61 samples). Prominent peaks in different textile fibre spectra were primarily found in the fingerprint region (1800–450 cm−1). To simplify analysis, the spectral data was reduced to principal components, and sample discrimination was performed using Python's PyCaret package. Multiple machine learning algorithms were explored for differentiating fibre samples, and the most effective one was selected for further validation. This study demonstrates the feasibility of developing an ATR-FTIR database for additional textile fibre samples, aiding in the detection of unknown or suspect fibres in the future.