Optical imaging of micro-droplets of dried saliva for oral squamous cell carcinoma diagnosis

Oral cancer, the sixth most common worldwide, is often diagnosed at an advanced stage, impacting patient survival and mortality. Liquid biopsy offers the potential for cancer diagnosis, enabling dynamic tumor monitoring and disease surveillance. Here we validates a novel diagnostic approach using optical images of dried micro-droplets (volume of one µ l) of saliva samples on glass and platinum substrates, employing Logistic Regression and Support Vector Machine (SVM) models. For each model, accuracy, sensitivity, specificity, and area under the ROC curve were calculated. Our findings indicated that optical density and surface area (SA) obtained from optical images of microdroplets are suitable paramters of discriminating oral cavity squamous cell carcinoma and health individuals. SVM models demonstrate impressive accuracy of 88.10% on glass and 95.00% on Pt substrates, ensuring robust and accurate detection of oral cancer based on these salient features.