Application of Improved Attention Mechanism Method in Thyroid Nodule Discrimination
This paper expounds the integration of the channel attention mechanism module into the discrimination of thyroid nodules.The framework is built on the DDTI public thyroid ultrasound dataset to distinguish between benign and malignant regions of interest on thyroid ultrasound images.The channel attention mechanism is introduced to correct the weight of the network's feature maps,effectively improving the accuracy of discrimination.The correct diagnosis rate of benign and malignant thyroid nodules on public datasets is 87％,which has great potential in clinical applications.
NETL
NSTL
万方数据
