Key points are not available for this paper at this time.
Deep learning-based object detection techniques have been developed, and they can help medical practitioners to identify and diagnose chest abnormalities. However, algorithms for chest anomaly detection are constrained by the imbalanced number of label categories and the irregular shape of anomalous regions with obscure feature boundaries.
Sun et al. (Tue,) studied this question.