Focal epilepsy has recently been associated with remote white matter damage, including reduced white matter volume. Longitudinal assessment of these white matter changes, in relation to functional mechanisms and consequences, may be ideally done by in vivo neuroimaging in well-controlled experimental animal models. We assessed whether advanced machine learning algorithm models could accurately detect volumetric changes in white matter from multiparametric MR images, longitudinally collected in a neocortical focal epilepsy rat model. We measured classification accuracy in two supervised segmentation models: i.e. the generalized linear model and the nonlinear random forest model-by comparing computed white matter probabilities with actual neuroanatomically identified white matter. We found excellent overall discriminatory power for both models. However, the random forest model demonstrated a superior goodness-of-fit calibration plot that was close to the ideal calibration line. Based on this model, we measured that total white matter volume increased in young adult control and epileptic rats over a period of 10 weeks, but the average white matter volume was significantly lower in the focal epilepsy group. Changes in gray matter volume were not significantly different between control and epileptic rats. Our results (1) indicate that recurrent spontaneous seizures have an adverse effect on global white matter growth and (2) show that individual whole brain white matter volume can be accurately determined using a combination of multiparametric MRI and supervised segmentation models, offering a powerful tool to assess white matter volume changes in preclinical studies of neurological disease.