Hierarchical classification models and Handheld NIR spectrometer to human blood stains identification on different floor tiles

Classification of bloodstains deposited at different times on floor tiles using hierarchical modelling and a handheld NIR spectrometer

“…The hierarchical classication models were built in the same way as has been detailed in the previous work. 7 Data analysis and pre-processing All data were analysed using MATLAB (R2010a 7.10.0.499, MathWorks) and the hierarchical models, SIMCA and PLS-DA models were built using PLS_Toolbox (Eigenvector Research, Inc.). Following the procedure described by Fonseca et al, 7 the spectra of all the samples were cut to the 1000-1600 nm range due to the presence of noise or other physical effects observed previously for this equipment with this type of sample.…”

“…7 Data analysis and pre-processing All data were analysed using MATLAB (R2010a 7.10.0.499, MathWorks) and the hierarchical models, SIMCA and PLS-DA models were built using PLS_Toolbox (Eigenvector Research, Inc.). Following the procedure described by Fonseca et al, 7 the spectra of all the samples were cut to the 1000-1600 nm range due to the presence of noise or other physical effects observed previously for this equipment with this type of sample. Savitzky Golay 1st Derivative technique with 15 smoothing points and a 2nd degree polynomial was used with all the data.…”

“…These models work as a decision tree where different classication techniques are connected in series, separated by a decision rule. 7,8 In the study developed by Fonseca et al, 7 different hierarchical models were built using blood samples that had been deposited for six days. Two decision rules were established for the development of each model.…”

“…These were selected based on types of the oor tiles used on the original training set (CE1, CE3, CE5, PO1, PO3 and PO4) and prediction set (CE4, CE6 and PO5), as described in the methodology detailed in a previous work. 7 Table S2 † shows the distribution of spectra in the training set and the different types of hierarchical models that were built. Table S3 † shows the external validation spectra distribution used for each model.…”

“…[1][2][3][4] Recent studies have shown the usefulness of portable equipment to identify bloodstains at suspected crime scenes. [5][6][7] Many challenges must be overcome for a practical application, such as the diversity of surfaces where these stains are found, [8][9][10] the environmental conditions to which they have been exposed, 1,11 and the length of time since the stains were deposited. 12 Different approaches have tested the application of both vibrational and chemometric techniques to identify bloodstains.…”