Rainer Rainer Krumm Krumm, Abstract Co-Author: Nothing to Disclose

Ludger Feyen, Abstract Co-Author: Nothing to Disclose

Walter Leonhard Heindel MD, Abstract Co-Author: Nothing to Disclose

Andre Kemmling MD, Presenter: Nothing to Disclose

To evaluate a novel algorithm for fully automated user independent assessment of regional relative cerebral CT perfusion (CTP) parameters in acute ischemic stroke.

Acute stroke imaging (native CT and dynamic CTP) was performed in 83 consecutive acute middle cerebral arterial strokes on a 128-slice CT scanner (Somatom Definition AS+, Siemens Medical Solutions). Brain perfusion maps were generated using a dedicated CTP workstation (Siemens syngo VE36A). Regional CTP values were obtain by a fully automated routine: (1) Robust linear affine image registration of CTP maps to MNI-152 space regardless of brain coverage was accomplished by two steps (FLIRT5.5). First, rigid body transformation of thresholded native CT (0 to 75 HU) to the anatomical average CT perfusion image was performed followed by inversion of the transformation matrix (CTPtoCT). Secondly, the matrix CTPtoCT and the transformation matrix CTtoMNI based on affine registration (12 DOF) of thresholded native CT to MNI were concatenated (CTPtoMNI). The matrix CTPtoMNI was applied to perfusion maps (cerebral blood-flow[CBF], -volume[CBV], time to peak[TTP]). (2) A probabilistic atlas in MNI space covering 68 structures (Harvard Center for Morphometric Analysis) was used to calculate regional brain perfusion. Perfusion parameters for all brain voxels weighted by the anatomical probability value of a region were calculated. (3) An error correction term was introduced to exclude non-significant voxels defined by cerebrospinal fluid and vessels. (4) Relative CT perfusion values were calculated (regional perfusion of ischemic hemisphere divided by the contra-lateral side). Automated relative perfusion values were compared to selected manually traced perfusion values for reference.

Linear registration of perfusion maps to MNI-152 space was robust and reliable. Automated and manually obtained relative regional brain perfusion values were not significantly different (p < 0.01). The average difference between automated and manually traced regional perfusion parameters was less than 4.3% (SD 3.1%).

The presented algorithm is robust and reliable for automated calculation of regional perfusion parameters.

The study targets the need for comprehensive automated user independent assessment of CT perfusion imaging and alleviates the need for manual identification of ischemia in perfusion maps.

Krumm, R, Feyen, L, Heindel, W, Kemmling, A, Fully Automated Assessment of Relative Cerebral CT Perfusion Parameters in Acute Ischemic Stroke.  Radiological Society of North America 2011 Scientific Assembly and Annual Meeting, November 26 - December 2, 2011 ,Chicago IL. http://archive.rsna.org/2011/11016980.html