Data for: Classification of benign-malignant thyroid nodules based on hyperspectral technology

We propose a rapid diagnostic method for benign and malignant thyroid nodules based on hyperspectral technology to address the issue of insufficient diagnostic efficiency in thyroid cancer during surgery. Firstly, through the self-developed thyroid nodule hyperspectral collection system, a large number of diverse thyroid nodule samples were obtained. These thyroid nodule samples were collected through the hyperspectral collection system during thyroidectomy surgery, providing a foundation for subsequent diagnosis. We propose a benign and malignant classification method based on hyperspectral data blocks of thyroid nodules to better meet clinical needs. Meanwhile, using 3D CNN and VGG networks, we designed a neural network algorithm for classifying three-dimensional hyperspectral cubes. The classification accuracy of benign and malignant samples reached 84.63%. Overall, we have effectively classified the benign and malignant thyroid nodules using a collection system and data.