Desenvolvimento de plataforma biofotônica de diagnóstico salivar para o carcinoma oral de células escamosas acoplada com algoritmos de inteligência artificial

• Autor(a) principal: Bellorin Nuñez, Daniela Sarai
• Data de Publicação: 2021
• Tipo de documento: Dissertação
• Idioma: por
• Título da fonte: Repositório Institucional da UFU
• Texto Completo: https://repositorio.ufu.br/handle/123456789/34309; http://doi.org/10.14393/ufu.di.2022.5306
• topic: Câncer oral; Oral cancer; carcinoma espinocelular; saliva; diagnostico; ATR-FTIR; biomarcador; squamous cell carcinoma; diagnosis; biomarker; CNPQ::CIENCIAS DA SAUDE::MEDICINA::CLINICA MEDICA::CANCEROLOGIA; Ciências médicas; Oncologia; Carcinoma de células escamosas

Introduction: The diagnostic methods for many types of cancer continue be the same used in recent years, to invasive and uncomfortable for patients, today we have advanced and a lot of innovative research for development tools with less waste production, lower costs, no invasive and reagent-free. Objective: In this study, we evaluate the potential of ATR-FTIR associated with artificial intelligence algorithms to discriminate OSCC from matched healthy controls. Materials and methods: The study included 39 OCSS patients and 27 healthy controls with no history of any type of cancer, all, saliva was collected directly from the oral cavity with plastic tube after after rinsing mouths with water and allowing the manufacturer instructions, subsequently they were placed on ice and then stored, later processed in the ATR-FTIR, the salivary profile was analysed by ATR-FTIR spectroscopy and the vibrational modes were evaluated for diagnostic capability by the ROC curve. Results: The lipidic band areas of 2894-2834 cm-1 and 2948-2896 cm-1 were higher in saliva of OCSS patients compared with the controls. These potential markers were further evaluated for detection of OCSS. The ROC curve was performed to predict a threshold value for both infrared biomarkers. The area under the curve (AUC) of both 2894-2834 cm-1 and 2948-2896 cm-1 regions was significative, indicating these parameters as potential biomarkers to discriminate OCSS and non-OCSS patients. Besides, two algorithms obtained by Random Forest and Support Vector Machine analysis were also useful to discriminate OSCC patients than healthy subjects with moderate accuracy. Conclusion: The present study indicated that machine learning algorithms and the salivary lipid vibrational modes 2894-2834 cm-1 and 2948-2896 cm-1 of saliva can be used to discriminate patients with OCSS than controls using photonic platforms in medical and dental environments.