Ischemic stroke is one of the major causes of mortality and accounts for most of adult disabilities happening worldwide. Early detection of ischemic stroke is of great importance since prolonged interruption of cerebral blood flow causes irreversible damage to brain tissues.This will lead to permanent disabilities or even death.Stroke diagnosis is mainly done using imaging systems. CT scan and MRI play a significant role in CVA detection. Non-enhanced CT is often the first radiologic examination performed in case of suspicion of stroke. Compared to MRI, brain imaging with CT is more accessible, less expensive and quicker. Nevertheless the signs of ischemia in few hours after onset are quite subtle on CT or practically invisible. Contrastis an important factor in subjective evaluation of image quality. In present thesis, we concentrated on contrast enhancement of brain CT images so that subtle signs of ischemia become more visible to radiologist. To this end numerous methods of contrast enhancement were evaluated. In medical image processing, multi-scale transforms are being used for different purposes. These transforms are able to extract image details and edges in different size and scales. So they provide a flexible framework for image analysis and modification. For example, by appropriate amplification of specific sub-bands’ coefficients using linear or non-linear mapping functions, we are able to emphasize on desired image details with specific size. This is our general approach in brain CT image enhancement. These transforms are completely characterized by their corresponding set of basis functions and this set can be redundant or not. By allowing redundancy, it is possible to enrich the set of basis functions so that the transform is more efficient in capturing some signal behavior. In addition, redundant transforms are generally more flexible and easier to design. In applications such as denoising, enhancement and contour detection a redundant transform can significantly outperform a non-redundant one. These redundant transforms are often translation-invariant. This feature is of unique importance in many image processing tasks. In this thesis, we used four different redundant multi-scale transforms to enhance brain CT images:Redundant Dyadic Wavelet Transform (RDWT), Laplacian Pyramid (LP), Nonsubsampled Pyramid (), and Nonsubsampled Contourlet Transfrom (NSCT). Except Laplacian Pyramid, the other three transforms are also translation invariant. At the end we show that by using these transforms and proper selection of sub-bands and mapping functions, it is possible to suitably enhance brain CT images so that ischemic sensitive brain tissues e.g. gray matter, basal ganglia and ... Keywords: stroke – ischemia – contrast enhancement – CT scan – Wavelet – NSWT – NSCT – Laplacian Pyramid