Machine-learning-assisted waveguide scattering microscopy for the immunological detection of bovine brucellosis: A proof concept

Autor(es):Mostrar menos -

Fonsaca, Jessica E. S. ; Teixeira, Wanderson S. R. ; Tieppo, Bianca ; Raitz, Cesar ; Rehan, Mohd ; Gerosa, Rodrigo M. ; Megid, Jane ; Appolinario, Camila M. ; Salles, Maiara O. ; Saito, Lucia A. M. ; Vale, Daniella L. ; Grasseschi, Daniel ; de Matos, Christiano J. S.

• Tipo de documento: Artigo Científico
• Fonte: SENSORS AND ACTUATORS B-CHEMICAL; v. 432, p. 10-pg., 2025-06-01.

Resumo

Photonic biosensors based on optical waveguides are at the forefront of biosensing technology, offering exceptional sensitivity and robustness. This study presents a proof-of-concept for detecting Brucella abortus antibodies in bovine serum using He-Ne laser-excited integrated optical waveguides. The antigen-antibody interactions in positive samples resulted in agglutination, forming scattering spots on the light-coupled waveguide, while no such spots were observed in negative samples. These spots were imaged over time using a microscope-coupled camera, generating 718 images that were processed to create a dataset for an artificial neural network (ANN). The ANN accurately distinguished between positive and negative samples, achieving 98.6 % accuracy, 98.7 % precision, and 98.7 % recall, with only a 1.4 % loss. This method detected bacterial antibodies in real animal samples within 20 min, using just 100 mu L of reagents without requiring prior waveguide surface modification for antigen immobilization. Combining light scattering-based sensing protocols in photonic waveguides with machine learning tools offers a promising pathway for revolutionizing infectious disease diagnostics. (AU)

• Processo FAPESP: 23/04830-0 - Biofuncionalização de biosensores fotônicos de guias de onda para diagnóstico de sepsis
• Beneficiário: Jéssica Eliza Silva Fonsaca
• Modalidade de apoio: Bolsas no Exterior - Estágio de Pesquisa - Pós-Doutorado
• Processo FAPESP: 22/07892-3 - Caracterização de dispositivos integrados não lineares
• Beneficiário: Mohd Rehan
• Modalidade de apoio: Bolsas no Brasil - Pós-Doutorado
• Processo FAPESP: 18/25339-4 - Dispositivos fotônicos integrados
• Beneficiário: Newton Cesario Frateschi
• Modalidade de apoio: Auxílio à Pesquisa - Temático
• Processo FAPESP: 20/13288-6 - Materiais bidimensionais e seus derivados para biossensoriamento fotônico de patógenos
• Beneficiário: Jéssica Eliza Silva Fonsaca
• Modalidade de apoio: Bolsas no Brasil - Pós-Doutorado