An MRI-based multi-objective radiomics model predicts lymph node status in patients with rectal cancer

Area of Science:

• Oncology
• Radiology
• Medical Imaging

Background:

• Lymph node (LN) metastasis is a critical prognostic factor in rectal cancer.
• Accurate pre-operative staging is essential for effective treatment planning.
• Radiomics offers potential for non-invasive assessment of tumor characteristics.

Purpose of the Study:

• To develop and validate a multi-objective radiomics model using pre-operative magnetic resonance imaging (MRI) to enhance the diagnostic accuracy of lymph node metastasis in rectal cancer.
• To compare the performance of the radiomics model against subjective radiologist assessments and a combined approach.

Main Methods:

• A cohort of 91 rectal cancer patients undergoing pre-operative MRI was analyzed.
• 1409 radiomic features were extracted from T2WI LN images.
• An iterative multi-objective immune algorithm (IMIA) was employed for feature selection and model building.
• Performance was evaluated using receiver operating characteristic (ROC) curves.

Main Results:

• The solitary radiomic model achieved 89.81% sensitivity, 82.57% specificity, and 87.77% accuracy.
• A combined model integrating radiomic and radiologist features demonstrated superior performance with 92.23% sensitivity, 84.69% specificity, and 89.88% accuracy (AUC of 0.94).
• Subjective radiologist analysis showed lower accuracy (78.12%).

Conclusions:

• The multi-objective radiomics model, particularly the combined approach, significantly improves the pre-operative prediction of lymph node status in rectal cancer.
• This T2WI-based radiomics model shows high utility for enhancing diagnostic accuracy in rectal cancer staging.