International Journal of Radiation Research

en Application of dual-parameter MRI-based radiomics in the differentiation of prostate cancer and benign prostatic hyperplasia: Diagnostic efficacy and predictive modeling Radiation Biology Radiation Biology تحقيق بديع Original Research <div style="text-align: justify;">Background: Prostate cancer (PCa) and benign prostatic hyperplasia (BPH) present with similar clinical symptoms, particularly in patients with borderline prostate-specific antigen (PSA) levels (4–10 ng/mL), making accurate diagnosis challenging. MRI-based radiomics enables non-invasive extraction of quantitative imaging features that may aid in differentiating these conditions. Materials and Methods: In this retrospective study, 150 patients (56 PCa, 94 BPH) underwent prostate MRI including T2-weighted imaging (T2WI) and diffusion-weighted imaging (DWI). Radiomics features were extracted from manually segmented lesions using PyRadiomics. Logistic regression was used for feature selection and model construction. Three predictive models were developed based on T2WI, DWI, and combined T2WI+DWI features. Model performance was assessed using receiver operating characteristic (ROC) analysis, evaluating area under the curve (AUC), sensitivity, and specificity on training and validation sets. Results: The combined T2WI+DWI model showed the best diagnostic performance with an AUC of 0.942 in the validation set, sensitivity of 0.821, and specificity of 1.000. This outperformed model based on T2WI or DWI alone, as well as the clinical model using PSA and prostate volume. Conclusion: Dual-parameter MRI-based radiomics enhances the non-invasive differentiation of PCa from BPH. The combined T2WI and DWI model offers superior diagnostic accuracy and may reduce unnecessary biopsies in patients with indeterminate PSA levels.</div> Prostatic neoplasms, radiomics, diffusion magnetic resonance imaging, computer-assisted, machine learning. 907 912 http://ijrr.com/browse.php?a_code=A-10-1-1414&slc_lang=en&sid=1 Y. Hu 7900319475328460032340 7900319475328460032340 No Department of Radiology,The First Affiliated Hospital of Chongqing Medical University, Chongqing, China C. Hou 7900319475328460032341 7900319475328460032341 No Department of Oncology, The First Hospital of Neijiang, Neijiang, Sichuan, China M. Wen mingwen_dc@163.com 7900319475328460032342 7900319475328460032342 Yes Department of Radiology,The First Affiliated Hospital of Chongqing Medical University, Chongqing, China