3D Reconstruction of Damage Caused by a Stroke using Magnetic Resonance Imaging Processing

Cerebrovascular accidents (stroke) can cause neurological problems, transient or permanent disability, and even death. In cases where disability is present, it is useful to be able to obtain specific information about the damage caused to the brain. One way to quantify the specific characteristics of the damage caused by stroke is to perform a three-dimensional reconstruction. It is for this reason that this paper proposes a method to carry out this procedure. Initially, the process used is based on automatic segmentation of the lesion in each of the images or slices that are part of the brain Magnetic Resonance (MR) sequence. The segmented regions are used together for lesion volume reconstruction. This volume is displayed on an interface, along with its location within the skull, and its specific characteristics. The results obtained comprise an interface showing data of interest about the lesion and its reconstruction. Also, in the lesion segmentation process, a favorable Dice-Sorense coefficient index of 0.75 is obtained with respect to the images annotated by experts. The lesion characteristics estimated in the implemented interface, such as location and volume, have an error of less than 30% with respect to the annotations of the database used, showing that the information provided by the proposed method is quite useful for the specialist's understanding of the brain damage caused by the stroke.