Elastic scattering spectroscopy for intraoperative oral cancer mucosal margin guidance: Initial results from a 104 patient cohort

Abstract

Objective

To assess Elastic Scattering Spectroscopy (ESS) classification accuracy of benign vs malignant tissue obtained during intra-operative oral cancer resection.

Methods

The study comprised 104 patients with a biopsy positive for oral cancer (N = 85) or dysplasia (N = 19) who were scheduled to undergo surgical excision. ESS measurements were obtained intraoperatively on and immediately adjacent to the lesion within the planned resection margin prior to excision, and on contralateral normal-site control tissue. Two-millimeter biopsies were obtained from tumor and margin tissue. All measurements were evaluated using Leave One Person Out (LOPO) AI-assisted statistical algorithms. Three analyses evaluated ESS diagnostic accuracy: one at the sample level, one at the pooled sample patient level, and one using only diagnostically variable biopsy co-registered margin samples. Statistical analyses included sensitivity, specificity, negative predictive value (NPV), positive predictive value (PPV), and Area Under the Receiver Operating Characteristic Curve (AUC-ROC).

Results

Diagnostic accuracy at the sample level yielded sensitivity = 82 %, specificity = 84 %, and AUC = 0.91. Pooling samples within each patient yielded sensitivity = 94 %, specificity = 87 %, and AUC = 0.95. Sample level diagnostic accuracy at the margin yielded sensitivity = 76 %, specificity = 50 %, and AUC = 0.70, but prioritizing sensitivity, yielded a sensitivity = 90 %, specificity = 30 %, with AUC = 0.70.

Conclusion

The ESS device demonstrated high sensitivity and appropriate specificity when differentiating benign from malignant tissue. Discriminant ability increased when samples were pooled within patients, informing future protocols for evaluating intraoperative ESS measures. These data are very promising and support the contention that ESS could be a valuable adjunct tool that facilitates comprehensive and efficient assessment of surgical margins.