Optofluidic label-free SERS platform for rapid bacteria detection in serum

• Published in: Sensors and actuators. B, Chemical Chemical, 2019-12, Vol.300, p.126907, Article 126907
• Main Authors: Hunter, Robert, Sohi, Ali Najafi, Khatoon, Zohra, Berthiaume, Vincent R., Alarcon, Emilio I., Godin, Michel, Anis, Hanan
• Format: Article
• Language: English
• Subjects: Algorithms; Antibiotics; Bacteria; Crystal fibers; Hospital acquired infection; Label-free detection; Machine learning; Microfluidics; Nanoparticles; Nosocomial infections; Optofluidic; Photonic crystals; Silver; Support vector machines; Surface enhanced Raman scattering
• ISSN: 0925-4005; 1873-3077
• DOI: 10.1016/j.snb.2019.126907

[Display omitted] •Surfaced enhanced Raman in hollow crystal fiber provides bulk 400-fold enhancement.•Fiber can be regenerated repeatedly with appropriate flushing fluid.•Counting bacteria spectral events can measure 4 CFU/mL in plasma.•Machine learning algorithm can discriminate between species with >90% accuracy. The prevalence of hospital acquired infections and antibiotic resistant pathogens necessitates the development of bacteria sensing systems that do not require sample amplification via conventional cell culturing, which can be prohibitively time-consuming. To meet this need, we designed an optofluidic Raman detection platform which utilized a microfluidic driven hollow-core photonic crystal fiber, which in combination with silver nanoparticles, provides a large enhancement to the Raman signal. By confining both light and cells within this fiber, spectral events generated by the flowing cells facilitates a novel method of cell counting to simultaneously quantify and qualify infections. Counting is performed automatically by a genetically optimized support vector machine learning algorithm that was previously developed by our group. The microfluidic system can be regenerated multiple times, and allows for online detection of planktonic bacteria to levels as low as 4 CFU/mL in 15 min. This compares favourably to other methods currently under development such as qPCR and biosensing techniques. Furthermore, Raman spectral differences between bacteria allow for inherent multiplexed detection in serum, by adding another layer to the learning algorithm. Further development of this device has promising potential as a rapid point-of-care system for infection management in the clinic.