Objective:To construct a bi-parameter radiomics model of magnetic resonance based on radiomics features of T2 fat suppression weighted images of magnetic resonance and diffusion-weighted apparent diffusion coefficient mapping,and to study the diagnostic value of the bi-parameter radiomics model for prostate cancer with clinical significance. Method:The imaging and pathological data of 93 patients with prostate cancer,who were confirmed at the First People's Hospital of Tianshui from January 2021 to February 2024,were retrospectively analyzed. They were divided into training set (65 cases) and testing set (28 cases) by using 7:3 principle of random number grouping method. The magnetic resonance T2 weighted fat suppression images and apparent diffusion coefficient mapping based on diffusion weighted imaging were preprocessed,and the regions of interest (ROI) were delineated. 1197 radiomics features were extracted. Machine learning support vector machine method was used to construct a bi-parameter omics model,which combined T2 weighted fat suppression images and apparent diffusion coefficient mapping based on diffusion weighted imaging. And then,the efficiency of this bi-parameter omics model of diagnosing prostate cancer with clinical significance was evaluated. Result:The area under curve (AUC) values of receiver operating characteristic (ROC) curve of training group and testing group based on the MRI bi-parameter model were respectively 0.928 and 0.894 in diagnosing prostate cancer with clinical significance. The sensitivity and specificity of that were respectively 84.2％ and 95.7％. Conclusion:The constructed bi-parameter radiomics model through training and testing has higher diagnostic value for prostate cancer with clinical significance,and it is relatively objective and accurate,which can provide reliable basis for clinical diagnosis and treatment of prostate cancer with clinical significance.
Prostate cancerMagnetic resonance imaging (MRI)RadiomicsDiagnostic value
维普
万方数据
