Automatic identification method and system for retinopathy treatment mode of premature infant

The invention relates to an automatic identification method and system for a retinopathy treatment mode of a premature infant, and the method comprises the following steps: S10, constructing a plurality of binary classifier models, a ResNet-18 model and a DenseNet-121 model are adopted in the method; s20, performing ROP screening on the target image by adopting a DenseNet-121 model; and S30, carrying out ROP identification and treatment prediction on the target image by adopting a ResNet-18 model. The model of the method is developed on the basis of a single training image, and predictions are generated on the image level. By adopting the method, the retinopathy treatment mode of the premature infant can be automatically identified, and the identification result is accurate. 本发明涉及一种早产儿视网膜病变治疗方式的自动识别方法及系统，识别方法包括以下步骤：S10、构建多个二元分类器模型，包括；ResNet-18模型和DenseNet-121模型；S20、采用DenseNet-121模型对目标图像进行ROP筛选；S30、采用ResNet-18模型对目标图像进行ROP识别和治疗预测。本发明方法的模型是基于单个训练图像开发的，它在图像级别上生成预测。采用本发明方法，可以实现早产儿视网膜病变治疗方式的自动识别，且识别结果准确。