Machine learning models based on MRI predict the expression levels of Galectin - 9 expression in rectal cancer
Yuhui Liu1, Shengtao Weng1, Dandan Wang2
1School of Medicine, Shaoxing University, Shaoxing, China.
Objective:
To investigate the utility of a machine learning model based on MRI radiomics in predicting the expression of Galectin-9 in rectal cancer.
Materials And Methods:
MRI images and clinical information of patients with locally advanced rectal cancer from Shaoxing People's Hospital from January 2019 to September 2024 were retrospectively analyzed. The patients were randomly divided into a training set and a test set at a ratio of 7:3. Radiological features were extracted from the regions of interest (ROI) of T2-weighted images (T2WI) and apparent diffusion coefficient (ADC) maps. The least absolute shrinkage and selection operator (LASSO) was used for feature selection. Three classifiers, namely extreme gradient boosting (XGBoost), support vector machine (SVM), and K-nearest neighbor classification (KNN), were used to construct T2WI, ADC, and combined models. The performance of the models was evaluated using 10-fold cross-validation and bootstrap validation (1000 iterations) .
Results:
The extreme gradient boosting (XGBoost) algorithm showed high comprehensive performance. In the training set, the AUC of the T2WI, ADC, and combined models was 0.886 (95%CI: 0.821-0.951), 0.730 (95%CI: 0.654-0.806), and 0.887 (95%CI: 0.823-0.951), respectively. In the test set, the AUC was 0.774 (95%CI: 0.689-0.859), 0.706 (95%CI: 0.615-0.797), and 0.808 (95%CI: 0.725-0.891), respectively.
Conclusion:
The machine learning model established based on multimodal MRI parameters can potentially facilitating personalized therapeutic stratification predict the expression of Galectin-9 in rectal cancer.
