Chest X-Ray Quality Assessment Method With Medical Domain Knowledge Fusion

Digital X-ray radiography is widely used in clinical diagnosis. High quality chest conducive to the accurate diagnosis of diseases by clinicians. However, assessment images mainly depends on subjective evaluation doctors, and results are influenced skill level experience evaluators, involving many issues such as heavy workload various uncertain factors judgment. In this paper, we propose a method that combines image-text contrastive learning with medical domain knowledge fusion. Based pretraining model from text-image pairs, large-scale real diagnostic report text information fused, fine-tuned achieve cross transfer learning. While improving prediction accuracy algorithm, cost massive sample data annotation avoided. The local visual patch features aligned multiple ensure contain more fine-grained image information. Theoretical analysis experimental show algorithm based fusion triplet graph multi-modal has achieved good performance terms accuracy. addition, proposed paper can be easily extended complete other tasks multi-lesion segmentation disease progression prediction.