Inflammatory idiopathic myositis is a debilitating inflammatory muscle condition. Diagnosis relies on a battery of tests, but monitoring of disease severity can be challenging. We present a novel machine learning approach to classifying tissues using multi-parametric analysis of routine MRI sequences. A logistic regression model was trained to predict tissue type based on T1 and STIR signal intensity and 10-fold cross-validated. The system attained 93.8% sensitivity and 96.9% specificity overall (ROC area 0.991). Testing of this model showed a low level of ostensible muscle inflammation in 9/11 asymptomatic controls – likely due to misclassification of vessels.