Contrast enhancement for brain MRI images via genetic algorithm-based dual cut histogram equalization

Abstract Contrast enhancement technique is a prevalent problem in aiding interpretation of Magnetic Resonance Imaging (MRI) images. A low contrast MRI image may prove to be problematic in identifying medical problems such as brain tumor in the image, either in visual identification or by automated algorithm such as machine learning. This study proposes a novel enhancement technique for Brain MRI images named Genetically Adapted Dual Cut Histogram Equalization (GADCHE). The main contribution of the paper is to provide technique to enhance the contrast of the image while simultaneously maintaining the structure of the image. The proposed method eliminates the extreme values of the histogram and partitions the histogram into three distinct groups and apply asymmetric transformation to them. The Genetic Algorithm side handles the optimization of the histogram’s cutting points to achieve optimum results. The experiments show that the proposed method gives promising results in terms of contrast enhancement across three datasets with average Effective Measure of Enhancement (EME) score of 263.25, Effective Measure of Enhancement Entropy (EMEE) score of 5.51, Logarithmic Michelson Contrast Measure (AME) score of 1.196, Logarithmic Michelson Contrast Measure by Entropy (AMEE) score of 0.0067 and Peak Signal to Ratio (PSNR) score of 30.06. The structural integrity of the resulting images are also shown to be intact with Structural Similarity Index (SSIM) score of 0.74 and Absolute Mean Brightness Error (AMBE) score of 9.75. In terms of contrast entropy, the proposed method is shown to perform other techniques, with competing results in terms of structural integrity.