Traumatic brain injuries represent a significant cause of disability and mortality globally. While survivors may experience substantial recovery in the first weeks or months after the traumatic incident, many encounter long-term challenges related to the cognitive, motor and psychosocial consequences of the injury. Diffusion-weighted brain imaging is a fundamental tool for understanding the pathophysiology of traumatic brain injury. These images provide a detailed view of the microstructure of brain tissue and facilitate the extraction of quantitative metrics that reflect structural integrity and brain connectivity. The present study aims to conduct a comprehensive statistical analysis of DWI data obtained from patients with traumatic brain injury in both the chronic and acute phases six months after the traumatic event, including healthy controls for comparison. In particular, our focus is on the analysis of metrics derived from brain graphs extracted from pre-processed DWI images. The goal is to identify early patients with acute and chronic traumatic injuries in comparison to healthy patients, who might show signs of cognitive impairment, by analysing brain connectivity metrics as possible predictive signals. Subsequently, we intend to employ eXplainable Artificial Intelligence (XAI) techniques to further our understanding of the factors influencing patients’ prognosis and recovery.