[Identification of osteoid and chondroid matrix mineralization in primary bone tumors using a deep learning fusion model based on CT and clinical features: a multi-center retrospective study].

• DOI: 10.12122/j.issn.1673-4254.2024.12.18
• Publication Date: 2024-12-20
• Authors: C Liu; Qingqing Zou; Menghong Wang; Qian Yang; Liwen Song; Zixiao Lu; Qianjin Feng; Yinghua Zhao
• Keywords: Osteoid; Primary bone; Bone matrix

We retrospectively collected CT scan data from 276 patients with pathologically confirmed primary bone tumors 4 medical centers in Guangdong Province between January, 2010 and August, 2021. A convolutional neural network (CNN) was employed as the deep learning architecture. The optimal baseline model (R-Net) determined through transfer learning, an optimized (S-Net) obtained algorithmic improvements. Multivariate logistic regression analysis used to screen clinical features such sex, age, mineralization location, pathological fractures, which were then connected imaging construct fusion (SC-Net). diagnostic performance of SC-Net machine models compared radiologists' diagnoses, their classification evaluated using area under receiver operating characteristic curve (AUC) F1 score.