Deep learning-based detection system for multiclass lesions on chest radiographs: comparison with observer readings

In this study, the authors sought to investigate the feasibility of a deep learning-based detection (DLD) system for multiclass legions on chest radiograph. The study included almost 16,000 chest radiographs collected from two tertiary hospitals, which were then used to develop a DLD system. The authors found that the DLD system showed potential for detections of lesions and pattern classification.

Key points

• The DLD system was feasible for detection with pattern classification of multiclass lesions on chest radiograph.
• The DLD system had high performance of image-wise classification as normal or abnormal chest radiographs (AUROC, 0.985) and showed especially high specificity (99.0%).
• In lesion-wise detection of multiclass lesions, the DLD system outperformed all 9 observers (FOM, 0.962 vs. 0.886; p < 0.001).

Article: Deep learning-based detection system for multiclass lesions on chest radiographs: comparison with observer readings

Authors: Sohee Park, Sang Min Lee, Kyung Hee Lee, Kyu-Hwan Jung, Woong Bae, Jooae Choe, Joon Beom Seo

Written by

European Radiology