Muscle MR Radiomics for Evaluation of Idiopathic Inflammatory Myopathies

This study aimed to establish a prediction model for Idiopathic Inflammatory Myopathies (IIM) based on the radiomics of vastus intermedius.

43 IIM patients and 48 control cases were analyzed in the retrospective study. By using the 3D slicer software, 107 radiomics features were obtained for each case. The selection of variables was performed using a two-sample t-test or the Mann-Whitney U test, followed by maximum correlation minimum redundancy (mRMR). Least Absolute Shrinkage and Selection Operator (LASSO) regression was used for dimension reduction. To assess the model performance, the receiver operating characteristic (ROC) curve was plotted.

The results showed that 67 features were varying between the IIM cases and control groups (p < 0.05). With the mRMR and LASSO methods, seven features were finally determined. Regarding the radiomics model's performance, it achieved an AUC of 0.983 in the validation data.

This study investigated microstructural differences in a specific muscle (varasimus intermedias) between Idiopathic Inflammatory Myopathy (IIM) patients and normal controls using radiomics. It identified 67 distinguishing radiomics features and successfully established a high-performance prediction model (AUC >0.9) using LASSO and mRMR for feature selection. These features likely reflect IIM-induced microstructural changes, such as inflammation and necrosis, typically visible on MRI as signal abnormalities. Radiomics thus offers a potential non-invasive alternative to muscle biopsy for assessing these changes.

Muscle MR radiomics is feasible for identifying idiopathic inflammatory myopathy and has the potential to serve as a non-invasive alternative to muscle biopsy.