Objective The aim of this study is to develop an open-source imaging database and an intelligent assisted diagnosis system for primary bone tumors.Methods Clinical information(gender,age,and location)and radiological da-ta(X-ray,CT,and MRI)from patients diagnosed with primary bone tumors at six healthcare facilities between January 2013 and December 2022 were retrospectively collected and inputted into the database,which was constructed on a cloud platform using B/S technology.A subset of the database's imaging data was utilized to develop and validate a diagnostic support sys-tem that classifies primary bone tumors into benign or malignant categories using deep learning algorithms applied to X-ray images.Results A user-friendly open-source imaging database for primary bone tumors has been successfully estab-lished,containing 2,127 cases;an auxiliary diagnostic system based on this databases X-ray images has been developed and integrated as a diagnostic module within the database,achieving an average AUC of 84.99％on the test set.Conclu-sion In this study,we constructed an open-source primary bone tumor image database and validated the development of a primary bone tumor-assisted diagnosis system based on it,which provides a powerful tool for bone tumor medical research and clinical practice.
Primary bone tumorMedical image databaseComputer-assisted diagnosisDeep learning
维普
万方数据
