A student researcher from my laboratory has been granted a national invention patent

Time: 2026-01-01 17:30:00

    The present invention pertains to the field of medical image technology and discloses a CT image classification method based on image segmentation and an RU classifier, as well as its applications. The method comprises the following specific steps: collecting CT images, performing preprocessing, and converting the preprocessed CT images into two-dimensional images to obtain a CT image dataset; constructing a segmentation network model comprising a symmetric encoder, decoder, and feature fusion layer; obtaining predicted segmentation maps of the CT image dataset using the trained segmentation network model; extracting radiomics features from the predicted segmentation maps; constructing an RU classifier, wherein the RU classifier includes parallel branches comprising a binary classification prediction model branch and an auxiliary function branch; acquiring and preprocessing CT images to be predicted, and sequentially processing them through the trained segmentation network model and RU classifier to generate survival classification results. The present invention addresses the limitations of existing survival prediction techniques, including lack of robustness, overfitting, and insufficient generalization capability, while offering the advantage of high prediction accuracy.