Abstract:OBJECTIVE: To establish a simple and feasible model of magnetic resonance imaging (MRI) for prediction of minimally conscious state in unconscious patients (>= 2 weeks) after severe traumatic brain injury (TBI).METHODS: MRI examinations were performed in 73 patients 4.5 weeks +/- 1.6 (range, 2-8 weeks) after TBI. Brain lesions on MRI, age, sex, cause of injury, Glasgow Coma Scale (GCS) score, and decompressive craniectomy were retrospectively analyzed. Outcome was assessed at 12 months from the onset of TBI.RESULTS: Of 73 patients, 39 were minimally conscious and 34 were unconscious at the endpoint. Binary logistic regression demonstrated that cause of injury (P = 0.036), GCS score (P=0.011), and lesions of the thalamus (P=0.002) and brainstem (P=0.012) shown on MRI were closely associated with the outcome of minimally conscious state. The overall correct prediction of the logistic model was 90.4%.CONCLUSIONS: The combination of MRI findings and other clinical data offers neurosurgeons substantial information about primary and secondary injuries of the patients with TBI, which allows a more accurate prediction of prognosis than a single GCS score or MRI findings alone. The regression model established in this study is simple and effective in predicting long-term unconscious state and minimally conscious state in patients after severe TBI.

The Role of Magnetic Resonance Imaging in the Prediction of Minimally Conscious State after Traumatic Brain Injury.