Stroke is one of the leading causes of death and disability in the world. A fast diagnosis is of paramount importance for enabling certain clinical interventions, such as thrombolysis or mechanical thrombectomy. Brain imaging is part of the diagnostic process, allowing physicians to assess the stroke lesion. However, their interpretation requires significant time and expertise, under high-pressure conditions, since every minute count for a better patient prognosis. Stroke segmentation is an area of great interest to clinical practice as it can help physicians to make informed time-critical decisions. Automated stroke segmentation can attenuate this process by providing both the location and the volume of the lesion. This work proposes a methodology for stroke segmentation using Computed Tomography (CT) scans, developed in the scope of the Ischemic Stroke Lesion Segmentation (ISLES)'24 challenge. We evaluate and compare two different architectures against other studies participating in the same challenge, using evaluation metrics such as Dice Similarity Coefficient (DSC), volume difference, F1-score and instance difference.