[Evaluation of the potential benefit of computer-aided diagnosis (CAD) for lung cancer screenings using photofluorography: analysis of an observer study].

To evaluate the potential benefit of computer-aided diagnosis (CAD) in lung cancer screenings using photofluorographic films, we performed an observer test with 12 radiologists. We used 60 photofluorographic films obtained from a lung cancer screening program in Yamaguchi Prefecture (30 contained cancerous nodules and 30 had no nodules). In these cases, our current automated detection scheme achieved a sensitivity of 80%, but yielded an average of 11 false-positives per image. The observer study consisted of three viewing conditions: 1) only the original image (single reading), 2) the original image and computer output obtained from the current CAD scheme (CAD 1), 3) the original image and computer output obtained from a simulated improved CAD scheme with the same 80% true-positive rate, but with an average of one false-positive per image (CAD 2). Compared with double reading using independent interpretations, which is based on a higher score between two single readings, CAD 2 was more sensitive in subtle cases. The specificity of CAD was superior to that of double reading. Although CAD 1 (Az = 0.805) was inferior to double reading (Az = 0.837) in terms of the ROC curve, CAD 2 (Az = 0.872) significantly improved the ROC curve and also significantly reduced observation time (p < 0.05). If the number of false positives can be reduced, computer-aided diagnosis may play an important role in lung cancer screening programs.