
Median Technologies has released trial results showing that its artificial intelligence (AI) algorithm performed well in characterizing malignant and benign lung nodules in patients receiving low-dose chest CT lung cancer screening.
The company's new iBiopsy computer-aided diagnosis (CADx) algorithm was trained on a dataset from 1,224 patients with 11,392 nodules from the National Lung Screening Trial and then tested on CT images from 472 patients with 4,216 nodules. The application yielded 95.2% sensitivity and 95.7% specificity for lung nodule characterization, as well as an area under the curve of 0.991, Median said.
The company is planning an additional large-scale study for an end-to-end digital lung cancer screening application that will provide both nodule detection and characterization. Results are expected in the fourth quarter.