Diffusion Kurtosis Imaging-Based Radiomics for Preoperative Prediction of Lymph Node Metastasis in Colon and Rectal...

Area of Science:

• Radiology and Imaging
• Oncology
• Medical Informatics

Background:

• Lymph node metastasis (LNM) is a critical prognostic factor in colorectal cancer (CRC).
• Accurate preoperative prediction of LNM is essential for effective treatment planning and patient management.
• Current imaging methods have limitations in precisely assessing LNM in CRC.

Purpose of the Study:

• To develop and validate a diffusion kurtosis imaging (DKI)-based clinical-radiomics model.
• To predict preoperative lymph node metastasis (LNM) in patients with colon and rectal cancer.
• To enhance the accuracy of LNM prediction beyond conventional imaging techniques.

Main Methods:

• Radiomic features were extracted from ADC, DKI_D, and DKI_K maps in 204 treatment-naive CRC patients.
• Feature selection was performed using least absolute shrinkage and selection operator regression.
• A combined clinical-radiomics model integrating DKI features and lymph node size was constructed and evaluated using ROC analysis.

Main Results:

• The combined DKI-based radiomics model achieved high predictive performance (AUC 0.939 training, 0.957 testing).
• This model significantly outperformed single-parameter models in predicting LNM.
• High performance was observed in both colon (AUC 0.980) and rectal (AUC 0.932) cancer subgroups.

Conclusions:

• The developed combined model offers a promising noninvasive tool for preoperative LNM prediction in colorectal cancer.
• This approach can aid in individualized risk stratification for CRC patients.
• The findings support the integration of DKI radiomics into clinical practice for improved treatment planning.