Diagnóstico de Leishmaniose Cutânea usando Espectroscopia FTIR e Aprendizagem de Máquina: um estudo em modelo animal

Abstract

Cutaneous Leishmaniasis (CL) is a polymorphic and spectral skin disease caused by Leishmania spp protozoan parasites. CL diagnosis is difficult, as usual methods are time consuming, expensive, and not very sensitive. Fourier transform infrared spectroscopy (FTIR) with machine learning (ML) algorithms has been explored as an alternative to achieve fast and accurate results for many disease diagnoses. Besides the good accuracy exhibited in many studies, the spectral differences between infected and non-infected groups are too subtle to be noticed. Since the sample set inhomogeneity (sex, age, diet, etc) usually carries great data variance and hinders a more detailed understanding of the spectral characteristics and immune responses, in this study, we explore this innovative methodology for CL diagnosis in an animal model study. Blood serum, skin lesions, and draining popliteal lymph node samples were obtained from BALB/C mice experimentally infected with Leishmania (Leishmania) amazonensis. FTIR and ML algorithms with reasonably good accuracy could differentiate between infected (CL group) and non-infected (control group) samples. The best overall accuracy (88.9%) was obtained using Principal Component Analysis (PCA) and Support Vector Machine (SVM) in the 4000-700 cm-1 range for blood serum samples. The accuracy obtained for skin lesions and popliteal lymph node samples was around 77.8%. The results suggest that a good performance of the prediction models is more associated with the data analysis than molecular composition differences in the samples.