A wireless handheld thyroid ultrasound artificial intelligence screening system and method

Abstract

This invention discloses a wireless handheld thyroid ultrasound artificial intelligence screening system and method. The system is integrated into a single handheld casing and includes an ultrasound probe, an imaging processing unit, a resource monitoring unit, a quality assessment and coverage estimation unit, a multi-level artificial intelligence inference unit, a screening decision and interaction unit, a wireless communication module, and a power supply module. The multi-level artificial intelligence inference unit deploys lightweight and medium-sized models locally and a high-precision model in the cloud. Each level of model outputs at least nodule candidate regions, malignancy confidence, and uncertainty indicators. The screening decision and interaction unit performs hierarchical inference, keyframe selection, and asynchronous uploading of weak network cache based on resource status and risk thresholds, enabling reliable screening and traceable final review in resource-constrained and unstable network environments.