Post-traumatic epilepsy (PTE) prediction is one of the greatest challenges in recent years. The probability of developing PTE is strongly connected with injury severity. Accordingly, having an automated alternative to clinician scoring, to measure injury severity, could be helpful to measure the progression of the disease in view of finding PTE biomarkers. Therefore, we have conducted a study aimed to evaluate if injury severity can be established from automatic analyses of MRI data in a way comparable to manual clinical scoring. We found a statistical association between morphological features and two clinical scores used to quantify injury severity.