New Research: A Breakthrough in Head and Neck Cancer Detection, Combining Blood and Saliva Spectra Boosts Diagnostic Accuracy to Over 90%
Scientific Reports
Our new study titled "Fused Raman spectroscopic analysis of blood and saliva delivers high accuracy for head and neck cancer diagnostics" by first author Hanna Koster has been published in the Nature journal Scientific Reports.
In this study we performed Raman spectroscopy of liquid biopsy samples (blood and saliva) collected from a large cohort of head and neck cancer patients and control subjects. Raman measurement of either plasma or saliva produced only moderate ability to distinguish cancer from control samples (~70% accuracy). However, when the two spectra for a given patient’s biofluids are stitched together prior to analytical data modeling, our models delivered highly accurate results (>90%). Our central finding is that this simple data augmentation method of stitching together Raman spectra across biofluids collected from each patient delivers a significant increase in diagnostic performance to the level needed for clinical use, something long promised but not delivered for vibrational spectroscopy.
These results are anticipated to advance the development of diagnostic applications of Raman spectroscopy, as well as other spectroscopy methods, optical and otherwise. And due to the growing interest in machine learning, the data augmentation method makes the potential impact of this work high across many other disciplines who are also working with sparse and/or small datasets!
This was a great collaborative effort by our team, working closely with Dr. Andrew Birkeland's clinical team at UC Davis Health.
As usual, our manuscript is fully open access and all raw data is published alongside the study here.
Download the pdf of the manuscript directly: