The invention is suitable for the technical field of medical
image processing and
artificial intelligence, and provides a
urinary system tumor
bone metastasis diagnosis method based on
deep learning and
bone positioning, and the method comprises the steps: collecting and preprocessing multi-
modal medical data containing PET / CT images; segmenting a skeleton region by adopting a
deep learning model, and realizing focus segmentation in the skeleton region; a lightweight skeleton segmentation network is constructed, and efficient and fine skeleton segmentation is realized through an
encoder-decoder structure; pET metabolic features, CT morphological features and texture features are extracted and fused, and a
radiomics feature matrix is constructed; carrying out
cancer species specificity analysis based on the
radiomics characteristic matrix, and constructing an intelligent diagnosis model by utilizing a
machine learning
algorithm; and finally, outputting a structured report containing the focus position, the type, the
cancer classification and the diagnosis confidence. According to the invention, through lightweight network design and multi-
modal feature fusion, the efficiency and accuracy of skeleton segmentation and focus diagnosis are significantly improved, and the method is suitable for clinical real-time auxiliary diagnosis.