Background and aim Prognosis of neonatal Arterial Ischaemic Stroke (AIS) seems to depend strongly on early MRI findings, while standard neurological examination provides limited prognostic value. The extent of a lesion as well as the brain areas involved appear to be the main outcome indicators. Moreover, classification according to arterial irrigation regions has proven useful. We aim to provide a tool for systematic vascular classification by means of a segmented 3D map with standard neonatal dimensions.
Methods A T1 structural MR image as well as a TOF angiography were deformed into a standardised neonatal brain. Brain structures and main arteries were then identified in this brain, and the corresponding arterial territories delineated with the help of published arterial region maps for adults. The standardised brain was then manually segmented, under supervision of three experts, and assembled into a 3-dimensional map.
Results Reliability of the resulting 3D map was assessed by automatically classifying a series of previously diagnosed AIS. Arterial region attribution mistakes were detected in radiology reports in some cases, supporting the need of this kind of tool. A test sample of 25 AIS with heterogeneous distribution and sizes was used to validate the original segmentation. Only minor corrections to territory boundaries were necessary.