The effect of head movement on CT perfusion summary maps: simulations with CT hybrid phantom data.

Head movement is common during CT brain perfusion (CTP) acquisition of patients with acute ischemic stroke. The effects of this movement on the accuracy of CTP analysis has not been studied previously. The purpose of this study was to quantify the effects of head movement on CTP analysis summary maps using simulated phantom data. A dynamic digital CTP phantom dataset of 25 time frames with a simulated infarct volume was generated. Head movement was simulated by specific translations and rotations of the phantom data. Summary maps from this transformed phantom data were compared to the original data using the volumetric dice similarity coefficient (DSC). DSC for both penumbra and core strongly decreased for rotation angles larger than approximately 1°, 2°, and 7° for, respectively, pitch, roll, and yaw. The accuracy is also sensitive for small translations in the z-direction only. Sudden movements introduced larger errors than gradual movement. These results indicate that CTP summary maps are sensitive to head movement, even for small rotations and translations. CTP scans with head movement larger than the presented values should be interpreted with extra care.

Head movement during CT brain perfusion acquisition of patients with suspected acute ischemic stroke.

OBJECTIVE: Computed Tomography Perfusion (CTP) is a promising tool to support treatment decision for acute ischemic stroke patients. However, head movement during acquisition may limit its applicability. Information of the extent of head motion is currently lacking. Our purpose is to qualitatively and quantitatively assess the extent of head movement during acquisition. METHODS: From 103 consecutive patients admitted with suspicion of acute ischemic stroke, head movement in 220 CTP datasets was qualitatively categorized by experts as none, minimal, moderate, or severe. The movement was quantified using 3D registration of CTP volume data with non-contrast CT of the same patient; yielding 6 movement parameters for each time frame. The movement categorization was correlated with National Institutes of Health Stroke Scale (NIHSS) score and baseline characteristic using multinomial logistic regression and student's t-test respectively. RESULTS: Moderate and severe head movement occurred in almost 25% (25/103) of all patients with acute ischemic stroke. The registration technique quantified head movement with mean rotation angle up to 3.6° and 14°, and mean translation up to 9.1mm and 22.6mm for datasets classified as moderate and severe respectively. The rotation was predominantly in the axial plane (yaw) and the main translation was in the scan direction. There was no statistically significant association between movement classification and NIHSS score and baseline characteristics. CONCLUSIONS: Moderate or severe head movement during CTP acquisition of acute stroke patients is quite common. The presented registration technique can be used to automatically quantify the movement during acquisition, which can assist identification of CTP datasets with excessive head movement.