In this paper, a two-dimensional truth-oriented brain model is proposed for imaging simulation of wearable stroke detection devices. Stroke detection based on ultra-wideband microwave imaging is a new detection method, which is portable, fast and safe. Wearable detection equipment is an important development direction of ultra-wideband microwave imaging for stroke. By analyzing the electromagnetic characteristics of the brain tissue at different frequencies, the electromagnetic parameters of the brain tissue were determined, and the real human brain MRI images were used for analysis and modeling. Compared to previous simple brain model, this model can more truly reflect the internal structure of human brain, and provide a more accurate experimental platform for various imaging simulation algorithms. Finally, the delay confocal algorithm was used to reconstruct the brain image. In the imaging results, the bleeding spots with a diameter of 6 mm in the brain is clearly identified.