Chinese researchers have discovered that MRI radiomics analysis derived from machine learning accurately predicted the differential degree (DD) and lymph node metastasis (LNM) of extrahepatic cholangiocarcinoma (CCA), according to a study published in BMC Cancer. 

“Radiomics, a recently introduced methodology, describes quantitative computerized algorithm-based feature extraction from traditional imaging data, including CT, MRI, or PET images,” the research team wrote. This approach allows for lesions to be quantified objectively, which differs from the standard approach of relying on radiologists for the visual interpretation of medical imaging.

Read more about CCA diagnosis 


Continue Reading

How successful would a radiomics model be in predicting DD and LNM of extrahepatic CCA? Answering that question was the objective of the study conducted by Tang et al. They recruited 100 patients diagnosed with extrahepatic CCA. From them, “a total of 1200 radiomics features were extracted from axial T1 weighted imaging (T1WI), T2-weighted imaging (T2WI), diffusion-weighted imaging (DWI), and apparent diffusion coefficient (ADC) images,” the researchers wrote. 

They then applied machine learning classifiers to the 1200 features selected. The results from the radiomics study demonstrated that the joint mutual information (JMI) feature selection method and bagging classifier achieved the best performance for DD prediction, with statistically significant sensitivity and specificity. 

As for LNM prediction, the radiomics study demonstrated that the minimum redundancy maximum relevance (MRMR) feature selection method and eXtreme Gradient Boosting classifier achieved the best performance, also with statistically significant sensitivity and specificity. 

This means radiomics can be useful in selecting the best features to identify the parameters of a disease.

“In conclusion, our MRI radiomics models based on optimal combinations of feature selection methods and machine learning classifiers demonstrate potentials in predictions for DD and LNM in extrahepatic CCA,” the researchers wrote. 

Reference

Tang Y, Yang CM, Su S, et al. Machine learning-based radiomics analysis for differentiation degree and lymphatic node metastasis of extrahepatic cholangiocarcinomaBMC Cancer. Published online November 24, 2021. doi:10.1186/s12885-021-08947-6