Diagnostic efficacy of MRI balanced steady-state free precession-based radiomics for benign and malignant diagnosis of peripheral lymph nodes in rectal cancer

[Objectives] To investigate the diagnostic efficacy of an MRI balanced steady-state free precession-based radiomics model for distinguishing between benign and malignant peripheral lymph nodes in rectal cancer. [Methods] A retrospective study was conducted on 288 rectal cancer patients treated at Lujiang Branch, Oriental Hospital, Tongji University from December 2020 to December 2023. All patients had histologically confirmed lymph node status, which was used to classify them into the benign lymph node group (n＝129) and the malignant lymph node group (n＝159). The patients were randomly divided into a training group (n＝201) and a validation group (n＝87) at a 7∶3 ratio. Radiomic features were extracted from both conventional T2WI sequences and balanced steady-state free precession sequences images by manually delineating region of interest around the lymph nodes. An effective radiomics model was constructed through data dimensionality reduction and feature selection. Receiver operating characteristic curve were plotted, and area under the curve, specificity, and sensitivity were calculated to quantify the diagnostic efficacy of the model in benign and malignant diagnosis of peripheral lymph nodes in rectal cancer. Decision curve analysis were used to evaluate the clinical utility of the model. [Results] The radiomics model based on balanced steady-state free precession sequences demonstrated superior diagnostic performance compared to conventional T2WI sequences. For the conventional T2WI sequences, the area under the curve, specificity, and sensitivity were 0.825, 0.726, and 0.755, respectively, in the training group, and 0.877, 0.829, and 0.832, respectively, in the validation group. For the balanced steady-state free precession sequences, the area under the curve, specificity, and sensitivity were 0.932, 0.950, and 0.868, respectively, in the training group, and 0.884, 0.902, and 0.845, respectively, in the validation group. The MRI balanced steady-state free precession-based radiomics model showed statistically significant higher diagnostic performance across all parameters and good calibration(P<0.05). [Conclusion] The MRI balanced steady-state free precession-based radiomics model can effectively predict the benignity and malignancy of peripheral lymph nodes in rectal cancer, providing valuable information for preoperative treatment decisions.