In a recent article published in the journal Scientific Reports, researchers presented the development of a novel binary biophotonic sensor designed for the rapid and sensitive detection of C-reactive protein (CRP) in human urine. By integrating advanced optical fiber technology with machine learning algorithms, the research aims to enhance the efficiency and accuracy of CRP measurement, ultimately improving patient outcomes.
Optical spectra measured in PBS after covering the microsphere surface with consecutive layers, forming the biolayer specific for CRP. Image Credit: https://www.nature.com/articles/s41598-024-67821-0
Background
C-Reactive Protein (CRP) is a critical biomarker used in clinical settings to assess inflammation and monitor various diseases. Its levels in the bloodstream can indicate acute inflammatory responses, making it a valuable tool for diagnosing conditions such as infections, autoimmune diseases, and chronic inflammatory disorders.
Traditional methods for CRP detection often involve complex procedures and lengthy analysis times, which can delay diagnosis and treatment. Moreover, conventional assays, such as enzyme-linked immunosorbent assays (ELISA), while effective, can be time-consuming and require specialized laboratory equipment.
Recent advancements in sensor technology have opened new avenues for rapid diagnostics. Optical sensors, particularly those utilizing fiber optics, offer advantages such as miniaturization, ease of use, and the potential for real-time monitoring. The integration of machine learning techniques further enhances the capability of these sensors by enabling sophisticated data analysis and classification.
The Current Study
In this study, researchers developed a binary biophotonic sensor by integrating a microsphere onto the tip of an optical fiber. The optical fiber was first prepared by cleaning and activating its surface to enhance the attachment of biofunctionalized materials. This preparation involved a series of chemical treatments, including immersion in dimethyl sulfoxide (DMSO) and acetone, followed by drying to create a reactive surface.
Next, a microsphere was attached to the end of the optical fiber and used as the sensing element. The microsphere was biofunctionalized through the sequential application of avidin and CRP-specific antibodies, which enabled the selective binding of CRP molecules in the sample.
To assess the sensor’s performance, a calibration curve was created using CRP solutions with concentrations ranging from 1.9 μg/L to 333 mg/L. Each measurement required a 10 μL sample, which was introduced to the sensor. The optical response was monitored using a spectrometer that captured near-infrared (NIR) spectra in the range of 1300–1600 nm, where both CRP and the surrounding structured water show significant absorption characteristics.
Data classification was conducted using machine learning algorithms, with a particular focus on the ExtraTrees classifier due to its robustness and efficiency in managing high-dimensional data. A dataset of 94 spectra was used for training and validation, enabling the evaluation of the sensor’s accuracy in classifying samples based on CRP levels. The entire process, from sample introduction to data analysis, was completed in under five minutes, highlighting the sensor’s potential for rapid clinical application.
Results and Discussion
The results showed that the developed sensor effectively detected CRP in both standardized solutions and real clinical samples from hospitalized patients. The calibration curve demonstrated a reliable response across the tested concentration range, confirming the sensor’s sensitivity and specificity. During the biofunctionalization process, optical spectra consistently showed a decrease in signal intensity, which correlated with the successful attachment of biomolecules to the microsphere surface.
In terms of classification performance, the ExtraTrees algorithm achieved high accuracy, correctly classifying 89 out of 94 spectra with only two false negatives. This performance highlights the potential of combining optical sensing with machine learning to enhance diagnostic capabilities.
The study also emphasized the importance of proper sensor maintenance. While the sensor head could be reused, it required thorough cleaning between samples to prevent cross-contamination. Immersion in ultrapure water was found to be effective for removing bound CRP, simplifying the cleaning process.
The sensor's ability to classify samples based on CRP levels in real time could significantly accelerate diagnosis and treatment decisions, especially in acute care settings. Furthermore, the integration of machine learning algorithms allows for ongoing improvements in classification accuracy as more data becomes available.
Conclusion
This study successfully developed a binary biophotonic sensor for the rapid and sensitive detection of CRP in human urine, showcasing its potential as a valuable tool for clinical diagnostics. The sensor leverages optical fiber technology and biofunctionalization techniques to achieve specific CRP detection, while the incorporation of machine learning algorithms enhances its classification accuracy.
The findings suggest that this innovative approach could significantly accelerate and improve the accuracy of diagnosing inflammatory conditions, thus benefiting patient care. Future research should focus on optimizing the sensor's performance, exploring its applicability to other biomarkers, and validating its effectiveness through larger clinical trials. Integrating such advanced diagnostic tools into routine clinical practice has the potential to revolutionize the management of inflammatory diseases and enhance overall healthcare outcomes.
Journal Reference
Cierpiak K., Wityk P., et al. (2024). C-reactive protein (CRP) evaluation in human urine using optical sensor supported by machine learning. Scientific Reports 14, 18854. DOI: 10.1038/s41598-024-67821-0, https://www.nature.com/articles/s41598-024-67821-0