Uso da espectroscopia FT-IR ATR e métodos de quimiometria em biofluidos para diagnósticos clínicos: efeito da diluição da amostra no diagnóstico da doença de Fabry

• Autor(a) principal: Leal, Leonardo Barbosa
• Data de Publicação: 2023
• Tipo de documento: Tese
• Idioma: por
• Título da fonte: Repositório Institucional da Universidade Federal do Espírito Santo (riUfes)
• Texto Completo: http://repositorio.ufes.br/handle/10/16935
• topic: Fisiologia; FTIR; Diagnóstico clínico; Quimiometria; Espectroscopia; subject.br-rjbn

Fourier Transform Infrared Spectroscopy (ATR-FTIR) emerges as a powerful tool to aid the area of clinical medicine. The FTIR technique provides accurate, real-time results for clinical/diagnostic screening tests. This study aimed to study the effect of plasma sample dilution on Fabry disease classification. Fabry disease (FD), a rare X cromossome disorder with alpha-galactosidase A (α-GAL-A) deficiency. The equipment used was the Agilent Cary 630 spectrometer, equipped with a diamond crystal, Total Attenuated Reflection (ATR) system. The range between 400 to 4000 cm-1 was analyzed with a resolution of 4cm-1 with 32 raster and background scans. Analyzes of volume, drying time and dilution were carried out. Once the best dilution point was located, plasma samples from patients with SCD were analyzed in their total and diluted form. Average spectra in triplicates were obtained and analyzed in separate regions, FG fingerprint (800-1800cm-1 ) and high HW wavenumber (2800-4000cm-1 ) and pre-processed by line vector normalization. Unsupervised analyzes were performed, such as principal component analysis (PCA) and Unsupervised Random Forest (URF) and classification analysis such as GA-LDA along with Monte Carlo Method with 500 models, Support Vector Machine (SVM), K-near Neighbor (KNN) and PLS-DA (Partial Least Squres – Discriminated Analysis) were used to obtain the results. The results suggested that the ideal drying time for plasma analysis is from 18 minutes and a volume of 10µl. For dilution, the best result was 25%. Both models built with diluted and undiluted samples were able to identify patients with SCD with specificity and sensitivity above 80%. Models built with undiluted samples were able to generate the best predictive model for PD. However, it is still necessary to investigate the effect of dilution in the making of predictive models created from other biological samples, such as, for example, urine.