Georg Homann, Presenter: Nothing to Disclose

Volker Vieth MD, Abstract Co-Author: Nothing to Disclose

Christian Juhra MD, Abstract Co-Author: Nothing to Disclose

Heindel Walter, Abstract Co-Author: Nothing to Disclose

Andre Kemmling MD, Abstract Co-Author: Nothing to Disclose

Quantitative CT imaging parameters are relevant for outcome prediction after traumatic brain injury. So far measurements of intracranial volume reserve as a possible factor in malignant brain edema and herniation has not been considered. The purpose of this study was to test if precise measures of intracranial volume reserve, volume of midline shift, and volume of traumatic brain lesions improves outcome prediction.

Trauma patients from the Münster Bicycle Study were selected based on presence of traumatic lesion on admission CT. Patients were dichotomized (endpoints malignant edema requiring craniectomy or death). Admission CT was used to determine volume of cerebrospinal fluid (CSF): after automated skull stripping and dynamic thresholding (HU range -10 to 20), CSF volume and intracranial volume (IV) were calculated from binary masks. Normalized CSF volume (nCSF) was defined as (volume of CSF)/IV. Traumatic brain lesion volumes (LV) targeted at 48h were outlined slice by slice with smart edge tracing, and normalized (nLV=LV/IV) (Analyze 10.0). As a indicator for herniation, the total volume of midline shift (MSV) outlined sliceby-slice was normalized to IV (nMSV = MSV/IV). The ratio of nLV/nCSF was correlated with nMSV. A voxel-wise probability map of nMSV, nLV, and nCSF in MNI-152 space was created. The cut-off value of nSV/nCSF for occurence of clinical malignant infarction was determined by ROC curve analysis.

A total of 75 cases were included. The ratio of nLV/nCSF significantly correlated with nMSV (correlation coefficient 0.76, p<0.001). The cut-off value for occurrence of craniectomy or death was 1.4 with a sensitivity of 90.1% and specificity of 89.3%. The voxel-wise probability map in the standardized MNI-152 brain allows direct visual estimation of nMSV with a priori knowledge of nCSF and nLV.  

We could show that there is a highly significant correlation between volume of midline shift and ratio of normalized traumatic lesion volme and CSF reserve volume. The model allows a rater independent prediction for occurrence of malignant edema requiring craniectomy or death after traumatic brain injury.  

Early decompression improves outcome in patients with malignant edema after traumatic brain injury. The presented model objectively identifies cases where a malignant course is highly probable.

Homann, G, Vieth, V, Juhra, C, Walter, H, Kemmling, A, Prediction of Fatal Outcome by a Rater Independent Model Based on Normalized Volume of Traumatic Brain Lesions and Subarachnoid Reserve Volume.  Radiological Society of North America 2012 Scientific Assembly and Annual Meeting, November 25 - November 30, 2012 ,Chicago IL. http://archive.rsna.org/2012/12038029.html