%0 Journal Article %A Jesse Kingg %A Steven Perrin %A Josh Knowland %A Jackson Kiser %A Spencer Bowen %T Development of a Quantitatively-Based 18F-FDG Infiltration Classifier of Topically Applied Sensor Readings %D 2019 %J Journal of Nuclear Medicine %P 63-63 %V 60 %N supplement 1 %X 63Objectives: To develop a new topical sensor-based 18F-FDG infiltration severity classifier, calibrated from quantitative PET image-derived measurements, and assess its performance on human subjects. Methods: A retrospective study examined patients whose 18F-FDG injections had been monitored as part of their standard workup for PET-CT imaging. Topical uncollimated gamma ray sensors were applied proximal to the injection site and on the same location on the opposing arm, and sensor readings acquired continuously during the uptake period starting just prior to tracer injection. The study population contains subjects with varying levels of infiltration identified using a qualitative assessment of sensor time-activity curves. Patients were imaged with their arms in the PET field-of-view. Total activity of the infiltration was quantified from PET images by taking the difference of summed activities from regions of interest placed on comparable sections of injection and reference arms. The image-derived activities were considered ground truth and used to calibrate and assess the quantification of topical sensor readings acquired at the time of PET imaging. The classifier utilizes the calibrated sensor readings as well as corrections for signal background corruption due to patient motion and liver uptake. In a blinded study, a radiologist qualitatively labeled the PET images for infiltration (none, minor, moderate, severe). The radiologist’s interpretations and topical sensor classifications were compared to the ground truth PET image-derived results. Results: Linear regression of topical sensor and image-derived tracer infiltration activity estimates showed high correlation, with R2=0.81. A total of 18 subject scans were cross-validated with the quantitatively-based classifier through a leave-one-out methodology. Specificity for binary infiltration classification (none and infiltrated) was 17% for the radiologist’s interpretations and 58% for the new topical sensor classification. Sensitivity was 100% for both. The physician tended to label infiltrations as more severe than the classifier. Conclusions: Preliminary findings suggest that this classifier, calibrated using quantitative static PET measurements, dramatically improves infiltration severity classification compared to qualitative image analysis. %U