Can near-infrared autofluorescence imaging be used for intraoperative confirmation of parathyroid tissue?

Abstract

Whether ex vivo autofluorescence (AF) imaging findings could be quantified to intraoperatively differentiate parathyroid tissue has not been reported. Our aim was to assess the ability of AF imaging to confirm parathyroid tissue during thyroidectomy and parathyroidectomy procedures. This was a retrospective Institutional Review Board-approved study. AF signals of specimens imaged ex vivo before submission to pathology during thyroidectomy and parathyroidectomy procedures were quantified. Using receiver operating characteristic (ROC) curves, optimal values for sensitivity/specificity to differentiate parathyroid tissue were calculated. 166 parathyroid and 217 non-parathyroid specimens were analyzed. With surgical drapes as background, the optimal normalized AF intensity threshold to predict parathyroid tissue on ROC curve analysis was 1.72 (86.1% sensitivity, 84.8% specificity, and area under the curve [AUC]: 0.919). The cutoff for 97% sensitivity was 1.31and for 99.1% specificity was 3.16. With Telfa as background, the optimal threshold to predict parathyroid tissue was 1.46 (88.4% sensitivity, 76.8% specificity, and AUC: 0.896). The threshold for 96.8% sensitivity was 1.18and for 98.8% specificity was 2.44. If thresholds for highest specificity were used, 40% of frozen sections to confirm parathyroid tissue could have been avoided. Quantified brightness analysis of ex vivo AF signals may have utility in intraoperative differentiation of parathyroid tissue for 40% of surgical specimens.