1) The study aimed to develop an automated routine for precise brain volume measurement in patients after aneurysmal subarachnoid hemorrhage (aSAH) using magnetic resonance (MR) images.
2) A k-nearest neighbor (kNN) classification approach was used, which requires manually segmented training data. The routine included brain extraction, inhomogeneity correction, and kNN classification to generate probability maps for different brain structures.
3) The routine was evaluated on MR images from 39 aSAH patients and 25 controls. Evaluation showed fractional similarity indices of 0.98, 0.93 and 0.92 for intracranial volume, total brain and lateral ventricles respectively, comparable to