Brain tissue segmentation for medical decision support systems

Automatic brain tissue segmentation on clinically acquired magnetic resonance image is a very challenging task due to the presence of intensity inhomogeneity, noise, and the complex anatomical structure of interest. Due to the existence of noise in clinical magnetic resonance brain images, various segmentation techniques suffer from low segmentation accuracy. Thus, to overcome the ambiguity caused by the above special effects, an enhanced fuzzy relaxation approach called fuzzy relaxation-based modified fuzzy c-means clustering algorithm is presented. In the proposed method, exposure-based sub-image fuzzy brightness adaptation algorithm is implemented for the enhancement of brain tissues, and it is followed by a modified fuzzy c-means clustering algorithm to segment the enhanced brain magnetic resonance image into white matter, gray matter and cerebrospinal fluid tissues. The proposed method is compared with other existing methods in terms of quantitative measures such as peak signal to noise ratio, discrete entropy, contrast improvement index, sensitivity, specificity, accuracy, jaccard similarity, and dice similarity coefficient. Experimental results demonstrate that the proposed method achieves a good trade-off between intensity inhomogeneity and noise. The proposed method conforms its success on brain tissue segmentation and provides extensive support to radiologists and clinical centers.