Multi-Class Brain Stroke CT Segmentation Dataset (Processed)

This dataset contains processed and anonymized non-contrast brain CT images and corresponding multi-class stroke lesion segmentation masks.The data are derived from a publicly available national stroke CT dataset provided through the Open Data Portal of the Ministry of Health of the Republic of Türkiye. The original dataset was prepared for the TEKNOFEST-2021 Artificial Intelligence in Healthcare competition and includes annotated brain CT scans with stroke labels. The dataset was originally introduced and described in: Koç U, Akçapınar Sezer E, Özkaya YA, Yarbay Y, Taydaş O, Ayyıldız VA, Kızıloğlu HA, Kesimal U, Çankaya İ, Beşler MS, Karakaş E, Karademir F, Sebik NB, Bahadır M, Sezer Ö, Yeşilyurt B, Varlı S, Akdoğan E, Ülgü MM, Birinci Ş. Artificial Intelligence in Healthcare Competition (TEKNOFEST-2021): Stroke Data Set. Eurasian Journal of Medicine, 2022;54(3):248–258. DOI: 10.5152/eurasianjmed.2022.22096. All patient-identifying information has been removed, and the data have been curated and reformatted for research use. The dataset consists of axial CT slices with pixel-wise annotations for ischemic and hemorrhagic stroke lesions.The images are provided for use as raw data in medical image segmentation and machine learning studies. The original dataset can be accessed at: https://acikveri.saglik.gov.tr This dataset supports the experiments reported in the manuscript "A Stable Diffusion–Based Framework for Multi-Class Stroke Lesion Segmentation in Brain CT".