VisionGate’s LuCED® test for lung cancer measures 934 features in each cell imaged in 3D by the Cell- CT™. Sensitivity and specificity, both exceeding 90%, have been reported (Wilbur, et. al., Cancer Cytopathology, 123:9,548-556). The relationship between sensitivity and an enumeration of bronchial epithelial cells (BECs) is shown in the SAT curve – for example, 1450 BECs yields a sensitivity of 90%. A normal result is assigned for cases with sufficient BECs and no detected abnormal cells.
LuCED AI was trained to identify cells with a cytologic diagnosis of atypia+. Accordingly, specificity for LuCED is measured for patients who are free of lung disease. Here, we investigate the rate of LuCED Submitted Abstract now in Review abnormal case reports for patients who are free of lung cancer but with or without non-cancer disease such as COPD. Two case-level ROC curves were produced by setting the required BEC count for a desired sensitivity and then recording the rate of LuCED abnormal reports for two control populations: 1. Free of lung disease, 2. With COPD or other non-cancer disease.
The study used 62 cases with lung disease and 37 cases without lung disease. The resulting case level ROC curves are shown below: