Computational method to measure vascular lesions in human brain digital images

Brain injury volume computation has clinical importance since stroke patient's recovery may be assessed with it. Tools currently employed in the clinical field can only compute injury volumes with qualitative and manual methods. The present work proposes a semiautomatic computational method for calculating these volumes. Users of the method are only required to select a single point comprising the injury from one of the slides of a series of medical images. This seed point is used by the proposed algorithm to compute the average and standard deviation of the gray scale magnitude of adjacent pixels which comprise the injury. The method was tested using magnetic resonance images and compared with the volume manually computed by experts using a medical image viewer software. The proposed method showed an absolute error of 7.8%, compared with the averaged volume computed by the experts. Therefore, it is possible to estimate volumes of brain injuries with the proposed semi-automatic algorithm. This method could help decrease the time invested by clinical personnel in computing brain lesions' volumes.