Intelligent diagnosis system for diabetic eye diseases based on multi-model cluster and causal enhanced LLM cooperation

Abstract

The application discloses a kind of intelligent diagnosis system of diabetic eye disease based on multi-model cluster and cause-effect enhanced LLM cooperation, and it is related to artificial intelligence medical auxiliary diagnosis technical field.The system includes modular detection cluster layer, microaneurysm, bleeding point, exudate, neovascular detection is carried out to input fundus image simultaneously, and the visual feature corresponding to diabetic eye disease is output;Medical text modal alignment layer, the diabetic eye disease visual feature output by modular detection cluster layer is converted into standardized medical description;Large language model diagnosis layer, according to the standardized medical description output by medical text modal alignment layer, combined with the construction of standardized medical description knowledge base system, the input fundus image is diagnosed and reasoned to diabetic eye disease, and structured diagnostic report is generated.The system has the advantages of high overall detection accuracy, knowledge evolution, diagnosis interpretability, output standardization and the like.

Images