Article Highlights
- A team of researchers from Southern University of Science and Technology in China developed a novel Raman spectral preprocessing scheme (RSPSSL) for biomedical applications.
- The RSPSSL method significantly reduces noise and errors in Raman spectra, improving accuracy in various biomedical applications, including cancer diagnosis and paraquat concentration prediction.
- This method exhibits versatility by enhancing Raman spectra from different applications, paving the way for label-free molecular imaging tools and high-throughput metabolomics profiling.
Biomedical applications involve the use of technology in conjunction with a live biological sample. They are often used in environmental, pharmaceutical, and materials science applications. One aspect to biomedical research researchers are looking into is how they can improve on existing technologies, whether that means innovating an existing device or creating a new one entirely.
A team of researchers from the Southern University of Science and Technology in Shenzhen, China, looked at how to improve Raman spectroscopy analysis in biomedical applications. To this end, lead researcher Gina Jinna Chen and her team developed and proposed a novel Raman spectral preprocessing scheme based on self-supervised learning (RSPSSL) (1). Their findings regarding the efficacy of their method were published in the journal Light: Science and Applications (1).
Raman spectroscopy is renowned for its molecular fingerprinting capability across various scientific domains. However, attributing vibration peaks amidst environmental, instrumental, and specimen noise poses a significant challenge (1). The RSPSSL method addresses this hurdle by intelligently preprocessing Raman spectra, removing statistical bias noise and other sample-related errors (1).
The researchers tested their RSPSSL, receiving positive results. Compared to established techniques, the RSPSSL method exhibited an 88% reduction in root mean square error (RMSE) and a 60% reduction in infinite norm (L∞) (1). Moreover, it showcased its versatility by enhancing various biomedical applications, including a 400% accuracy improvement in cancer diagnosis and significant improvements in paraquat concentration prediction (1).
