Radiomics for Differentiating Prostate-Specific Membrane Antigen (PSMA)-Positive and PSMA-Negative Lymph Nodes in Prostate Cancer: An Exploratory 68Ga-PSMA PET-CT Study.

Introduction Accurate lymph node assessment in prostate cancer remains clinically relevant for staging and treatment planning. Although 68Ga-prostate-specific membrane antigen (PSMA) PET-CT has improved nodal evaluation, quantitative imaging biomarkers may provide additional information beyond visual assessment. This exploratory study aimed to characterize CT-derived radiomic features extracted from PSMA-positive and PSMA-negative lymph nodes using 68Ga-PSMA PET/CT positivity as the imaging reference standard. Materials and methods This retrospective, cross-sectional study included 41 patients with prostate cancer who underwent 68Ga-PSMA PET-CT. A total of 414 lymph nodes were segmented, including 62 PSMA-positive and 352 PSMA-negative nodes. Radiomic feature extraction was performed using 3D Slicer and pyradiomics. Twenty-nine radiomic features were evaluated and categorized into shape and texture feature families, including gray level size zone matrix, gray level run length matrix, neighboring gray tone difference matrix, and gray level dependence matrix. The descriptive comparison and diagnostic performance of all evaluated radiomic features were reported. Statistical comparisons were performed between PSMA-positive and PSMA-negative lymph nodes, and receiver operating characteristic curve analysis was used to assess discriminatory performance. Results Several radiomic features differed significantly between PSMA-positive and PSMA-negative lymph nodes. The highest discriminatory performance was observed for long run high gray level emphasis, with an area under the curve of 0.763, sensitivity of 77.42%, and specificity of 67.90%. Other high-performing features included high gray level emphasis, with an area under the curve of 0.751; large dependence high gray level emphasis, with an area under the curve of 0.746; high gray level run emphasis, with an area under the curve of 0.744; short run high gray level emphasis, with an area under the curve of 0.731; and low gray level emphasis, with an area under the curve of 0.726. Conclusion Radiomic features extracted from 68Ga-PSMA PET-CT differed between PSMA-positive and PSMA-negative lymph nodes and demonstrated moderate discriminatory performance. The highest-performing features were predominantly texture-based metrics, suggesting that radiomic analysis may provide complementary quantitative information for lymph node characterization in prostate cancer. However, given the exploratory design, class imbalance, manual segmentation, lack of individual-node histopathological confirmation, and absence of external validation, these results should be interpreted as hypothesis-generating. Future prospective studies with standardized segmentation, histopathological correlation, and multivariable model validation are needed before clinical implementation.