Reconstructing traumatic brain injury loading conditions remains challenging in forensic biomechanics. This study presents a proof-of-concept inverse-analysis framework for identifying plausible head impacts from post-traumatic brain imaging. It was applied to a pedestrian struck by a car who sustained severe brain injury and a malar fracture. Cerebral microbleeds on T1-weighted MRI and regions of abnormal fractional anisotropy on diffusion tensor imaging served as injury markers. A partially personalized finite element head model was subjected to 120 impacts compatible with the facial fracture, varying velocity, orientation, and contact surface. Predicted regions of elevated maximum principal strain (>0.2) were compared with observed injury locations using the Matthews correlation coefficient (MCC). With T1-weighted lesions alone, the best simulation achieved an MCC of 0.11, with high specificity (0.84) but low sensitivity (0.50), reflecting the difficulty of reproducing sparse, dispersed microbleeds. Including regions of abnormal fractional anisotropy improved performance, yielding an MCC of 0.34 and a sensitivity of 0.884. The best scenarios involved impacts to the lower posterior maxillary region, near the mandible and zygomatic bone, at 4–7 m/s. These findings support lesion-informed inverse finite element modeling for forensic impact reconstruction and suggest that fractional anisotropy alterations may provide complementary information on brain deformation.