Nonenhanced computerized tomography (CT) exams were used to detect acute stroke by notification of hypodense area. Infarction perception improvement by data denoising and local contrast enhancement in multi-scale domain was proposed. The wavelet-based image processing method enhanced the subtlest signs of hypodensity, which were often invisible in standard CT scan review. Thus improved detection efficiency of perceptual ischemic changes was investigated. Data processing became more effective by initial segmentation of brain tissue and extraction of regions susceptible to tissue density changes. The new method was experimentally verified. Sensitivity of stroke diagnosis increased to 56.3% in comparison to 12.5% of standard CT scan preview.