Clinical Application of Computer-Aided Detection System for Pulmonary Nodules on Digital Chest Radiography / 中山大学学报(医学科学版)

ABSTRACT

[Objective] To observe the effect of computer-aided detection (CAD) system in improving lung nodule detection sensitivity and inter-observer variation.[Methods] 300 PA digital radiographs including 100 normal cases and 200 cases with pulmonary nodules confirmed by CT were enrolled.Two senior chest radiologists referenced CT images and marked the sizes and locations of all nodules with consensus as the gold standards.Four senior radiologists and four junior radiologists interpreted the digital chest radiographs independently without and with CADand recordtheir results.Pair t test and coefficient of variation (CV) was used to compare the difference of lung nodule detection sensitivity and inter-observer variation between withoutand with CAD.[Results] The mean lung nodule detection sensitivity of senior and junior radiologists withoutand with CAD were (41.1 ± 2.0)％,(28.0 ± 2.0)％ and (45.0 ± 1.8)％,(39.2 ± 0.9)％,respectively,statistical analysis showed there was statistically significant difference.Moreover,CV of all radiologists without and with CAD were 20.9％ and 8.1％.[Conclusion] Both lung nodule detection sensitivity and inter-observer variation of senior and junior radiologists can be improved by CAD.