Intelligent segmentation of MRI images with the help of genetic algorithm to extract brain tumor

Background and aims: Although MRI can accurately describe brain structures, tumor segmentationin medical images is difficult. The use of genetic algorithm in the processing of brain MRIimages can lead to accurate diagnosis of tumor from healthy tissue.Method Genetic algorithm as an evolutionary algorithm has a high ability to optimize image processingparameters. This algorithm was used in the MATLAB environment to optimize the fuzzyC-means clustering method in the segmentation of brain MRI images containing tumors. Then,with the help of thresholding methods and morphological operators, the final processing was doneon the image. Finally, to evaluate the proposed segmentation method, image evaluation criteriawere checked.Results: The average value obtained for the criteria of similarity, accuracy, sensitivity and validitywas calculated as ۰.۸۷, ۹۹.۴۳%, ۹۹.۳۷% and ۹۹.۴۳% respectively. which is favorable comparedto previous studies. The PSNR criterion value was also obtained within an acceptable range. Butthe MSE values were slightly enlarged, which requires further investigations.Conclusion: The proposed method provides an intelligent hybrid method for tumor detection andextraction from brain MRI images. This method does not need previous information for segmentationand also increases the speed of the detection process.

کلیدواژه ها:

Genetic algorithm ، image processing ، brain tumor intelligent diagnosis ، MRI ، MATLAB